5G QoS intelligent configuration method and device based on natural language processing

Abstract

The invention provides a 5G QoS intelligent configuration method and device based on natural language processing. The method comprises the following steps: receiving a natural language QoS configuration request of a user; analyzing the natural language QoS configuration request by using a natural language analysis model enhanced by domain knowledge, and outputting a structured QoS configuration parameter set conforming to a 3GPP (3rd Generation Partnership Project) protocol specification; verifying and processing the structured QoS configuration parameter set through cooperative interaction of a plurality of network elements in the 5G core network, and generating a final executable QoS strategy instruction; and generating a policy decision based on the finally executable QoS policy instruction and real-time network context information, and controlling a user plane function network element to execute the policy decision to complete QoS configuration. The method has the beneficial effects that the professional threshold and technical complexity of 5G QoS configuration are remarkably reduced, the configuration efficiency and accuracy are greatly improved, and network resources are allocated according to needs and dynamically optimized at the same time.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and communication technology, and in particular to a 5G QoS intelligent configuration method and device based on natural language processing. Background Technology

[0002] The development of fifth-generation mobile communication technology (5G) has driven the arrival of the era of the Internet of Things, expanding its network service targets from traditional public users to numerous vertical industries such as industrial control, intelligent driving, and telemedicine. These industry applications have placed differentiated and extreme demands on network performance, such as latency, bandwidth, and reliability. Quality of Service (QoS) configuration, as a core technology for ensuring service experience, is becoming increasingly important. In the 5G architecture, according to 3GPP standards, QoS configuration requires the coordinated interaction of multiple network elements, such as Network Open Function (NEF), Session Management Function (SMF), and Policy Control Function (PCF), involving a large number of specialized parameters such as 5QI, GBR, and ARP, with strict constraints between these parameters.

[0003] Currently, mainstream 5G QoS configuration methods heavily rely on network operators' in-depth understanding of 3GPP protocol specifications and manual operation. Configuration personnel not only need to accurately grasp the technical meaning of various parameters and their complex mapping rules (e.g., the correspondence between 5QI and packet delay budget (PDB)), but also need to be familiar with the interface protocols and call processes between various network elements. This manual-driven configuration model has revealed several significant shortcomings in practice: First, the configuration threshold is extremely high, making it difficult for non-telecommunications industry users to participate directly, hindering the rapid deployment of 5G networks; second, manual configuration is inefficient and prone to errors due to human negligence, making it difficult to meet the needs of rapid service launch and dynamic adjustment; third, existing methods lack scenario adaptability and cannot dynamically and intelligently optimize parameters based on real-time changes in network conditions (such as load and congestion) and diverse service requirements.

[0004] Therefore, a key technical bottleneck in existing technologies is how to lower the professional threshold for 5G QoS configuration, automate and intelligentize the configuration process, and improve the efficiency of network element collaboration and the dynamic adaptability of policy decisions. Solving this problem is crucial for the large-scale application and deep penetration of 5G networks across various industries. Summary of the Invention

[0005] The technical problem to be solved by this invention is: how to automatically and accurately convert non-professional natural language QoS requests into QoS policies that conform to 3GPP specifications and can be collaboratively executed by 5G core network elements, so as to reduce the configuration threshold and realize intelligent decision-making.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a 5G QoS intelligent configuration method based on natural language processing, comprising: Receive users' natural language QoS configuration requests; The natural language QoS configuration request is parsed using a domain knowledge-enhanced natural language parsing model, and a structured QoS configuration parameter set conforming to the 3GPP protocol specification is output. Through the collaborative interaction of multiple network elements in the 5G core network, the structured QoS configuration parameter set is verified and processed to generate the final executable QoS policy instruction. Based on the final executable QoS policy instructions and real-time network context information, a policy decision is generated, and the user plane function network elements are controlled to execute the policy decision to complete the QoS configuration.

[0007] Furthermore, the process of parsing the natural language QoS configuration request using a domain-knowledge-enhanced natural language parsing model includes: The natural language parsing model is initialized based on a large language model, and then fine-tuned using domain knowledge with 3GPP protocol texts and historical configuration cases. Construct a QoS domain knowledge graph, which includes parameter entities, network element entities, constraint relationships between parameters, and 3GPP mandatory rules; By using the fine-tuned natural language parsing model and the QoS domain knowledge graph, the fuzzy expressions in the natural language QoS configuration request are mapped to standardized QoS parameters.

[0008] Furthermore, during the parsing process, if there are logical contradictions among the identified parameters, automatic correction or a prompt message will be sent to the user.

[0009] Furthermore, the verification and processing of the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network includes: The network open function element performs field integrity and syntax verification on the received structured QoS configuration parameter set; Query the unified data management network element to verify user subscription data and application access permissions; Based on the predefined parameter mapping table and operator policies, the missing QoS parameters in the structured QoS configuration parameter set are supplemented, and parameter constraint verification is performed to ensure logical consistency.

[0010] Furthermore, the step of verifying and processing the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network also includes: Query the network storage function network element to obtain the load status and resource capacity of the target session management function network element; Based on the Network Slicing Service Level Protocol, evaluate whether the structured QoS configuration parameter set is within the capability range of the target slice; If an anomaly is detected during preprocessing, a standardized error code is returned to the requester.

[0011] Furthermore, the generation of policy decisions based on the final executable QoS policy instructions and real-time network context information includes: The policy control function network element integrates user subscription information, pre-configured policy rules, real-time network status and third-party application requirements, performs compliance verification, resource adaptation and policy conflict resolution, and generates policy decisions. The session management function network element converts the policy decision into a forwarding rule that can be executed by the user plane function network element.

[0012] Furthermore, after generating and executing the policy decision, it also includes: Through the session management function network element and the network open function network element, the final execution result of the QoS configuration is converted into natural language and fed back to the user.

[0013] Furthermore, after the user plane function network element executes the policy decision to complete the QoS configuration, a closed-loop optimization process is also included, specifically: Collect actual QoS indicator data of user plane function network elements; The actual QoS indicator data is compared with the target parameters of the QoS configuration. If the target is not met, the session management function network element dynamically adjusts the configuration parameters of the user plane function network element.

[0014] Furthermore, the 5G QoS intelligent configuration method based on natural language processing also includes a case study and optimization process, specifically including: Store the triplet data of successful QoS configuration cases in the case library; Based on the aforementioned case library, the mapping relationship from natural language to structured parameters is optimized; When a request similar to that in the case library is received, the optimal configuration in the case library should be reused first.

[0015] The present invention also provides a 5G QoS intelligent configuration device based on natural language processing, comprising: The configuration request receiving module is used to receive users' natural language QoS configuration requests; The configuration request parsing module is used to parse the natural language QoS configuration request using a domain knowledge-enhanced natural language parsing model and output a structured QoS configuration parameter set that conforms to the 3GPP protocol specifications. The policy instruction generation module is used to verify and process the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network, and generate the final executable QoS policy instruction. The decision execution module is used to generate policy decisions based on the final executable QoS policy instructions and real-time network context information, and control the user plane function network elements to execute the policy decisions to complete the QoS configuration.

[0016] The beneficial effects of this invention are as follows: By deeply integrating natural language processing technology with 5G core network standard processes, an end-to-end intelligent configuration channel from user intent to network execution is constructed. This channel can automatically and accurately convert vague service requirements proposed by non-professionals into standardized QoS parameters that comply with 3GPP specifications. Through a multi-network element collaborative mechanism, the parameters are intelligently verified, completed, and decided, ultimately driving user plane execution. This significantly reduces the professional threshold and technical complexity of 5G QoS configuration, greatly improves configuration efficiency and accuracy, and simultaneously realizes on-demand allocation and dynamic optimization of network resources. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the 5G QoS intelligent configuration method based on natural language processing according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the configuration request parsing process according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating the parameter verification process according to an embodiment of the present invention. Figure 4 This is a flowchart of parameter anomaly handling according to an embodiment of the present invention; Figure 5 This is a block diagram of a 5G QoS intelligent configuration device based on natural language processing, according to an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be noted that the descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0021] like Figure 1 As shown, an embodiment of the present invention is: a 5G QoS intelligent configuration method based on natural language processing, comprising the following steps: S10, Receive the user's natural language QoS configuration request; In this embodiment, users can be operators of an industrial IoT platform or dispatchers of an autonomous driving system. Without in-depth knowledge of 3GPP protocols, they can submit their network performance requirements in natural language via a graphical user interface or API. For example, a user could input: "Create a highly reliable link for the video quality inspection service of production line A, ensuring an upload bandwidth of no less than 50Mbps and a latency of less than 20ms." This step directly addresses the business scenario, converting complex professional configuration commands into intuitive business language descriptions, fundamentally lowering the user's barrier to entry.

[0022] S20. The natural language QoS configuration request is parsed using a domain knowledge-enhanced natural language parsing model, and a structured QoS configuration parameter set conforming to the 3GPP protocol specification is output. In one specific embodiment, such as Figure 2 As shown, in step S20, parsing the natural language QoS configuration request using a domain-knowledge-enhanced natural language parsing model includes: S21. Initialize the natural language parsing model based on the large language model, and fine-tune the natural language parsing model with domain knowledge using 3GPP protocol text and historical configuration cases; In this embodiment, a general-purpose large language model (such as DeekSeek-R1, LLaMA, etc.) is first selected as the foundation. Then, domain-specific training corpora are collected and constructed. These corpora include: original texts of protocols such as 3GPP TS 23.501 (system architecture) and TS29.552 (NEF interface), operator QoS configuration guidelines, and numerous successful historical configuration cases (such as autonomous driving - 5QI=2 - GBR=100Mbps). A domain adapter is used to fine-tune the base model, which can efficiently inject 5G QoS domain expertise while preserving the model's ability to understand general language. The fine-tuned model can accurately understand the specific meanings of vague expressions such as high reliability and low latency in the 5G context, laying the foundation for subsequent structured output. This process addresses the shortcomings of general-purpose large models in terms of domain-specific knowledge, ensuring the professional accuracy of the parsing.

[0023] S22. Construct a QoS domain knowledge graph, which includes parameter entities, network element entities, constraint relationships between parameters, and 3GPP mandatory rules. In this embodiment, the knowledge graph is the core component ensuring the correctness of the parameter logic. Its entity layer defines key concepts, such as parameter entities "5QI", "ARP", and "GBR", and network element entities "SMF", "PCF", and "NEF". The relationship layer defines constraints between these entities, such as: 5QI=1; PDB (Packet Delay Budget) ≤ 10ms; ARP priority values ​​range from 1 to 15, with smaller values ​​indicating higher priority. The rule layer stores mandatory rules, such as: GBR services must be configured with 5QI=1-4, as well as operator-defined rules, such as: a certain type of UPF device supports a maximum GBR of 1Gbps. This knowledge graph serves as an external knowledge base for the parsing model, working in conjunction with the model to verify and constrain the parameters output by the model, preventing the generation of configurations that violate protocols or device capabilities.

[0024] S23. Using the fine-tuned natural language parsing model and the QoS domain knowledge graph, the fuzzy expressions in the natural language QoS configuration request are mapped into standardized QoS parameters.

[0025] In this embodiment, the user request "guaranteeing a high-reliability link with an upload bandwidth of no less than 50Mbps and a latency of less than 20ms" is input into the finely tuned model. The model reasons under the constraints of a knowledge graph: "Guaranteed upload bandwidth" is identified as requiring the configuration of a GBR (Guaranteed Bit Rate) service, and "50Mbps" is assigned to gbrUl (uplink GBR); "latency less than 20ms" is mapped to the 5QI parameter. By querying the mapping relationship between 5QI and PDB in the knowledge graph, 5QI=3 (PDB=20ms) or 5QI=80 (PDB=10ms) are identified as candidates. Considering the high reliability requirement, the more reliable 5QI=3 is ultimately selected. Simultaneously, high reliability is mapped to a higher priority in ARP (Assignment and Reservation Priority), such as priorityLevel=5. Finally, the model outputs a structured JSON object containing fields such as {"5qi": 3, "arp": {"priorityLevel": 5}, "gbrUl": "50Mbps"}. This step enables the precise conversion from unstructured, fuzzy natural language to standardized, machine-readable structured data.

[0026] Furthermore, during the parsing process, if there are logical contradictions among the identified parameters, automatic correction or a prompt message will be sent to the user.

[0027] In this embodiment, logical verification is performed using knowledge graphs and models. For example, if a user requests "configure 5QI=1 for a web browsing service (default for voice calls)," the system will detect that the "web browsing" service type is severely mismatched with the typical application scenario of "5QI=1." In this case, the system can perform two operations: first, automatically correct it to a more suitable 5QI=9 (non-GBR data) based on the service type "web browsing"; second, generate a prompt, returning to the user "The 5QI=1 you specified is typically used for voice calls and does not match the web browsing service. We recommend using 5QI=9. Do you want to continue?". This mechanism greatly improves the system's robustness and user experience, avoiding configuration errors caused by improper user input.

[0028] S30. Through the collaborative interaction of multiple network elements in the 5G core network, the structured QoS configuration parameter set is verified and processed to generate the final executable QoS policy instruction. In one specific embodiment, such as Figure 3 As shown, in step S30, the verification and processing of the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network includes: S31. The network open function element performs field integrity and syntax verification on the received structured QoS configuration parameter set; In this embodiment, the NEF, acting as a gateway to external applications, first verifies the integrity of the request body according to the Nnef interface data model defined in the 3GPP TS 29.552 protocol. For example, it checks the existence of the required fields qosReference (QoS policy identifier) ​​and qosParameters (including 5QI, ARP, etc.). Simultaneously, it performs syntax validation to ensure the parameter types are correct, such as whether arp.priorityLevel is an integer between 1 and 15, and whether gbrUl is a string with units (e.g., "50 Mbps"). This step ensures that requests entering the core network are formatted correctly and legally, intercepting low-level syntax errors.

[0029] S32. Query the unified data management network element to verify user subscription data and application access permissions; In this embodiment, NEF queries the UDM via the Nudm interface to obtain the user's (identified by SUPI) subscription information, such as the maximum allowed 5QI tier and maximum aggregate uplink and downlink bandwidth. Simultaneously, NEF verifies the OAuth 2.0 access token provided by the external application and confirms, based on the pre-configured exposureControlPolicy, whether the application has permission to request the specified QoS resources. Furthermore, NEF anonymizes sensitive user identifiers (such as IMSI) to comply with the privacy and security specifications of 3GPP TS33.501. This step ensures the legitimacy and security of the configuration request from the user authorization and subscription perspectives.

[0030] S33. Based on the predefined parameter mapping table and operator policies, complete the missing QoS parameters in the structured QoS configuration parameter set, and perform parameter constraint verification to ensure logical consistency.

[0031] In this embodiment, the system has a built-in parameter mapping table and default policy. For example, a user may only specify the service type as autonomous driving without explicitly specifying 5QI. The system will automatically complete it to 5qi=2 based on the mapping table. If the user does not specify the QoS Flow Identifier (QFI), the system will automatically assign the default value qfi=63. Simultaneously, deep parameter constraint validation is performed. For example, if the request contains 5qi=9 (non-GBR service) but the gbrUl field is configured, the system will trigger correction logic to automatically remove the GBR-related fields. This step ensures that the final generated QoS policy instruction is complete and self-consistent in terms of parameters, compensating for any omissions that may occur during the parsing phase.

[0032] In one specific embodiment, such as Figure 4 As shown, in step S30, the verification and processing of the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network further includes: S34. Query the network storage function network element to obtain the load status and resource capacity of the target session management function network element; S35. Based on the Network Slice Service Level Protocol, evaluate whether the structured QoS configuration parameter set is within the capability range of the target slice; S36. If an anomaly is detected during preprocessing, a standardized error code is returned to the requester.

[0033] In this embodiment, the NEF queries the NRF through the Nnrf interface to discover the SMF instance that can serve the user session and obtain information such as its current CPU load and the number of sessions it carries. Simultaneously, based on the SLA (Service Level Agreement) of the network slice (eMBB slice) to which the session belongs, the NEF assesses whether the requested QoS parameters (e.g., 50Mbps GBR) are within the resource capacity of that slice. If the assessment finds that the target SMF load is too high or the requested bandwidth exceeds the slice limit, the NEF will terminate the process and return a standardized error code based on 3GPP TS 29.571, such as RESOURCE_LIMIT_EXCEEDED. This allows external applications to obtain a clear reason for the operation failure, rather than a vague network error. This step achieves resource pre-assessment based on real-time network status, avoiding the forwarding of unfulfillable requests to the core network decision layer and improving overall system efficiency.

[0034] S40. Generate a policy decision based on the final executable QoS policy instruction and real-time network context information, and control the user plane function network element to execute the policy decision to complete the QoS configuration.

[0035] In one specific embodiment, step S40, which involves generating a policy decision based on the final executable QoS policy instruction and real-time network context information, includes: S41. The policy control function network element integrates user subscription information, pre-configured policy rules, real-time network status and third-party application requirements, performs compliance verification, resource adaptation and policy conflict resolution, and generates policy decisions. In this embodiment, the SMF forwards the request carrying qosParameters to the PCF. The PCF then makes the highest-level policy decision. It integrates four information sources: 1) the user QoS subscription file obtained again from the UDM (as the final authority); 2) pre-configured SmPolicy rules, such as automatically reducing the ARP of eMBB services when congestion occurs in area X; 3) networkAreaInfo reported by the SMF, such as the UE's real-time location and current network slice load; 4) the original requests of third-party applications forwarded by the NEF. The PCF's execution logic is as follows: First, compliance verification is performed. If the user's subscription maximum GBR is 80Mbps, but the request is 100Mbps, it is directly rejected. Next, resource adaptation is performed. If the request is for 50Mbps but the current radio interface resources are scarce, the PCF may decide to downgrade its gbrUl to 30Mbps and mark qosDegradation=true. When there is a policy conflict, such as the user's subscription allowing high bandwidth, but the temporary operation and maintenance policy requires rate limiting, the execution priority is determined according to the policyPriority field. Ultimately, PCF generates a complete policy decision that includes billing rules (rating groups), usage monitoring rules (such as traffic threshold alarms), and event subscriptions (such as notifying applications when QoS cannot be guaranteed). This step is core to enabling refined and intelligent operation of 5G networks, ensuring that network behavior is always consistent with business strategies and real-time status.

[0036] S42. The session management function network element converts the policy decision into a forwarding rule that can be executed by the user plane function network element.

[0037] In this embodiment, after receiving the policy decision from the PCF, the SMF converts the qosProfile (QoS profile) into underlying data plane rules that the UPF can understand and execute. For example, it maps 5QI=3 to a specific DSCP tag, creates a dedicated QoS Flow for GBR traffic, and sends the corresponding Packet Inspection Rules (PDR) and Forwarding Action Rules (FAR) to the UPF in the N4 interface session establishment or modification request, instructing the UPF on how to identify, queue, and forward packets belonging to this QoS flow. Simultaneously, the SMF stores the smPolicyId associated with the current PDU session context for subsequent policy updates. This step completes the final conversion from control plane policy to user plane execution, making the QoS configuration truly effective.

[0038] Furthermore, after generating and executing the policy decision, it also includes: Through the session management function network element and the network open function network element, the final execution result of the QoS configuration is converted into natural language and fed back to the user.

[0039] In this embodiment, after configuration, SMF notifies NEF of the success result (including the final effective parameters, such as {"5qi": 3, "gbrUl": "50Mbps"}) or the reason for failure (such as insufficient resources, configuration failure) through the Nnef_QoSControl_Notify service. NEF then invokes its natural language generation capability to convert the structured result into user-friendly natural language, such as: Your video quality inspection service QoS policy has been successfully created, and the system has guaranteed you 50Mbps uplink bandwidth and a transmission latency of less than 20ms. This forms a complete interactive loop, greatly improving the experience for non-professional users.

[0040] Furthermore, after the user plane function network element executes the policy decision to complete the QoS configuration, a closed-loop optimization process is also included, specifically: S51. Collect actual QoS indicator data of user plane function network elements; S52. The actual QoS indicator data is compared with the target parameters of the QoS configuration. If the target is not achieved, the session management function network element dynamically adjusts the configuration parameters of the user plane function network element.

[0041] In this embodiment, the SMF periodically collects real-world QoS metrics from the UPF via the N4 interface, such as packet latency (e.g., measured at 15ms), jitter, and bandwidth utilization. The SMF compares the measured latency (15ms) with the configured target (PDB = 20ms for 5QI=3), finding that while it is within the target, the margin is small. To provide a better experience, the SMF can proactively initiate an N4 session modification request to dynamically adjust the buffer queue length of the corresponding QoS stream on the UPF, for example, reducing it from the default 50ms to 30ms, to further reduce latency. This dynamic tuning mechanism based on real-time data upgrades QoS assurance from static configuration to dynamic assurance, better addressing fluctuations in network conditions.

[0042] Furthermore, the 5G QoS intelligent configuration method based on natural language processing also includes a case study and optimization process, specifically including: S61. Store the triplet data of successful QoS configuration cases into the case library; S62. Based on the aforementioned case library, optimize the mapping relationship from natural language to structured parameters; S63. When a request similar to that in the case library is received, the optimal configuration in the case library shall be reused first.

[0043] In this embodiment, each successful configuration process, including its natural language request, structured parameters, and execution result (including optimized metrics collected by UPF), is stored as a triplet case in the case library. Based on this case library, optimization can be performed in two ways: First, based on reinforcement learning, for example, if the system discovers that a historical case mapping low latency to 5QI=2 had a significantly better measured latency (8ms) than a case mapping to 5QI=3 (15ms), it will automatically optimize the mapping strategy, favoring 5QI=2 in the future. Second, based on case matching, when a new request arrives to configure a low-latency link for AR teaching services, the system matches similar historical successful cases in the case library and can directly reuse their proven optimal configuration (e.g., 5QI=80, GBR=20Mbps), thus skipping complex parsing and decision-making chains and achieving millisecond-level configuration response. This process makes the system smarter with use, resulting in increasingly accurate and efficient configurations.

[0044] like Figure 5 As shown, this embodiment of the invention also provides a 5G QoS intelligent configuration device based on natural language processing, comprising: Configuration request receiving module 10 is used to receive users' natural language QoS configuration requests; The configuration request parsing module 20 is used to parse the natural language QoS configuration request using a natural language parsing model enhanced with domain knowledge, and output a structured QoS configuration parameter set that conforms to the 3GPP protocol specification. The policy instruction generation module 30 is used to verify and process the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network, and generate the final executable QoS policy instruction. The decision execution module 40 is used to generate a policy decision based on the final executable QoS policy instruction and real-time network context information, and control the user plane function network element to execute the policy decision to complete the QoS configuration.

[0045] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned 5G QoS intelligent configuration device based on natural language processing can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0046] The aforementioned 5G QoS intelligent configuration device based on natural language processing can be implemented as a computer program that can run on a computer device.

[0047] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A 5G QoS intelligent configuration method based on natural language processing, characterized in that, The method comprises the following steps: receiving a natural language QoS configuration request of a user; parsing the natural language QoS configuration request by using a natural language parsing model enhanced by domain knowledge, and outputting a structured QoS configuration parameter set conforming to a 3GPP protocol specification; verifying and processing the structured QoS configuration parameter set through the cooperative interaction of multiple network elements in a 5G core network, and generating a final executable QoS policy instruction; generating a policy decision based on the final executable QoS policy instruction and real-time network context information, and controlling a user plane function network element to execute the policy decision to complete QoS configuration. 2.The 5G QoS intelligent configuration method based on natural language processing of claim 1, wherein, The parsing of the natural language QoS configuration request by using the natural language parsing model enhanced by domain knowledge comprises the following steps: initializing a natural language parsing model based on a large language model, and fine-tuning the natural language parsing model by using 3GPP protocol texts and historical configuration cases; constructing a QoS domain knowledge graph, which comprises parameter entities, network element entities, constraint relationships between parameters, and 3GPP mandatory rules; mapping fuzzy expressions in the natural language QoS configuration request to standardized QoS parameters by using the fine-tuned natural language parsing model and the QoS domain knowledge graph. 3.The 5G QoS intelligent configuration method based on natural language processing of claim 2, wherein, In the parsing process, when there is a logical contradiction between the identified parameters, automatic correction or prompt information is sent to the user. 4.The 5G QoS intelligent configuration method based on natural language processing of claim 1, wherein, The verification and processing of the structured QoS configuration parameter set through the cooperative interaction of multiple network elements in the 5G core network comprises the following steps: performing field integrity and syntax verification on the received structured QoS configuration parameter set by a network exposure function network element; querying a unified data management network element to verify user subscription data and application access permissions; completing missing QoS parameters in the structured QoS configuration parameter set based on a pre-defined parameter mapping table and operator policies, and performing parameter constraint verification to ensure logical consistency.

5. The natural language processing based 5G QoS intelligent configuration method of claim 4, wherein, The verification and processing of the structured QoS configuration parameter set through the cooperative interaction of multiple network elements in the 5G core network further comprises the following steps: querying a network repository function network element to obtain the load state and resource capability of a target session management function network element; evaluating whether the structured QoS configuration parameter set is within the capability range of a target slice based on a network slice service level agreement; if an exception is detected in the preprocessing process, returning a standardized error code to the requester. 6.The natural language processing based 5G QoS intelligent configuration method according to claim 1, characterized in that, The generation of a policy decision based on the final executable QoS policy instruction and real-time network context information comprises the following steps: integrating user subscription information, pre-configured policy rules, real-time network state and third-party application requirements by a policy control function network element, performing compliance verification, resource adaptation and policy conflict resolution to generate a policy decision; converting the policy decision into a forwarding rule executable by a user plane function network element by a session management function network element.

7. The natural language processing based 5G QoS intelligent configuration method of claim 6, wherein, After the generation and execution of the policy decision, the following steps are further included: converting the final execution result of QoS configuration into natural language feedback to the user by a session management function network element and a network exposure function network element. 8.The 5G QoS intelligent configuration method based on natural language processing of claim 6, wherein, After the control of the user plane function network element to execute the policy decision to complete the QoS configuration, a closed-loop optimization process is further included, specifically including: Collecting actual QoS index data of the user plane function network element; Comparing the actual QoS index data with target parameters of the QoS configuration, if the target is not reached, the session management function network element dynamically adjusts the configuration parameters of the user plane function network element. 9.The natural language processing based 5G QoS intelligent configuration method according to claim 1, wherein, Further including a case learning and optimization process, specifically including: Storing the three tuple data of the successful QoS configuration cases into a case library; Optimizing the mapping relationship from natural language to structured parameters based on the case library; When receiving a similar request in the case library, the optimal configuration in the case library is preferentially reused. 10.A 5G QoS intelligent configuration device based on natural language processing, characterized in that, Including: A configuration request receiving module for receiving a natural language QoS configuration request of a user; A configuration request analysis module for analyzing the natural language QoS configuration request by using a natural language analysis model enhanced by domain knowledge, and outputting a structured QoS configuration parameter set conforming to the 3GPP protocol specification; A policy instruction generation module for checking and processing the structured QoS configuration parameter set through the collaborative interaction of multiple network elements in the 5G core network, and generating a final executable QoS policy instruction; A decision execution module for generating a policy decision based on the final executable QoS policy instruction and real-time network context information, and controlling the user plane function network element to execute the policy decision to complete the QoS configuration.

Images