CN-122027562-A - Network service quality intelligent management and control system based on real-time intention recognition and perception

CN122027562ACN 122027562 ACN122027562 ACN 122027562ACN-122027562-A

Abstract

The invention discloses a network service quality intelligent management and control system based on real-time intention recognition and perception, and belongs to the technical field of digital information transmission. The invention solves the technical problems of intent identification and perception separation, static analysis and difficult end-to-end coordination in the existing network management and control. The system comprises a real-time intention recognition module, a multi-dimensional perception fusion module, an intelligent management and control decision module, a cross-layer collaborative execution module and an end-to-end service quality guarantee, wherein the real-time intention recognition module is used for carrying out implicit intention mining on real-time communication demands of a user according to user behavior data and service flow characteristics to obtain real-time user intention, the multi-dimensional perception fusion module is used for carrying out fusion modeling on multi-dimensional perception data by utilizing the real-time user intention and network state information to generate intention perception joint characteristics, the intelligent management and control decision module is used for carrying out deep reinforcement learning decision on network resources and service flows by virtue of the intention perception joint characteristics to obtain a dynamic management and control strategy, and the cross-layer collaborative execution module is used for completing cross-layer collaborative scheduling on a terminal, access, bearing, a core network and application by virtue of the dynamic management and control strategy.

Inventors

XU GANG
ZHENG XUHUA

Assignees

杭州华思通信技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (9)

1. The intelligent network service quality management and control system based on real-time intention recognition and perception is characterized by comprising the following modules: The real-time intention recognition module is used for implementing implicit intention mining on the real-time communication requirements of the user according to the user behavior data and the service flow characteristics to obtain real-time user intention; the multi-dimensional perception fusion module is used for performing fusion modeling on multi-dimensional perception data by utilizing the real-time user intention and the network state information to generate intention perception joint characteristics; The intelligent management and control decision module is used for developing deep reinforcement learning decisions on network resources and service flows by means of the intention perception joint characteristics to obtain a dynamic management and control strategy; And the cross-layer cooperative execution module is used for completing cross-layer cooperative scheduling for the application of the terminal access bearing core network by means of the dynamic management and control strategy to form end-to-end service quality guarantee.
2. The intelligent network quality of service management and control system based on real-time intent recognition and perception according to claim 1, wherein the means for obtaining the intent of the real-time intent recognition module comprises: According to the user behavior data and the business flow characteristics, adopting a time sequence behavior analysis model to implement joint modeling on the user historical behavior sequence and the current session characteristics, and obtaining user intention transfer points and implicit intention; By means of text input or voice instruction of a user interface, semantic analysis and intention key parameter extraction are carried out on the input by using a natural language processing technology, and explicit intention is generated.
3. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as recited in claim 1, wherein the source of data of the real-time intent recognition module comprises: Extracting a user movement track and an operation habit by means of the user terminal positioning data and an application use log to obtain a user geographic position change track, a foreground application switching record, an interactive operation time interval and a historical preference label; And performing deep packet detection and flow analysis on the service by utilizing the network flow data to obtain a service type identification result, a flow burst mode and QoS requirement parameters.
4. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as claimed in claim 1, wherein the awareness dimension of the multidimensional awareness fusion module comprises: Real-time monitoring and statistics are carried out on network performance through network measurement data, and real-time bandwidth occupancy rate, one-way delay and round trip delay statistics, continuous packet loss event record and link quality fluctuation index are obtained; And analyzing the application layer protocol and the flow characteristics according to the service flow data to obtain an application layer protocol identifier, a data packet size distribution characteristic, a flow duration and a service session stage mark.
5. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as claimed in claim 1, wherein the fusion mechanism of the multidimensional awareness fusion module comprises: According to the real-time user intention, performing dynamic adjustment on the sampling frequency and the precision of each sensing dimension to form a high-density sampling configuration aiming at the high-priority intention-related sensing dimension; By means of network state, service characteristics, user behaviors and environmental context data, the multi-source heterogeneous data is subjected to association modeling by adopting a graph neural network, and intention perception joint characteristics are generated through characteristic intersection and attention mechanisms.
6. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as recited in claim 1, wherein the decision algorithm of the intelligent management and control decision module includes: real-time decision making is carried out on network resources and service flows by utilizing a deep Q network and combining experience playback and a target network mechanism, so that a dynamic management and control strategy for maximizing long-term cumulative rewards is obtained; And carrying out joint optimization on the real-time user intention achievement degree and the overall network resource efficiency by means of a multi-objective optimization model of weighted summation and pareto front, and outputting a dynamic management and control strategy.
7. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as recited in claim 1, wherein the policy output of the intelligent management and control decision module includes: According to the intention perception joint characteristics, differential QoS guarantee strategy formulation is respectively executed on video service, data service and real-time call service, and a dynamic management and control strategy comprising high-bandwidth low-delay path allocation, elastic bandwidth adjustment and priority queue setting is obtained; By means of the intention achievement level prediction model, an estimation is performed on the quality change within a future time window, obtaining a resource reservation or path switching preconditioning strategy for proactively triggering before quality degradation.
8. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as recited in claim 1, wherein the cooperative range of the cross-layer cooperative execution module includes: According to the dynamic management and control strategy, executing the adjustment of the transmitting power and the modulation and coding scheme on the terminal side, executing the dynamic allocation of the resource blocks and the scheduling priority to the access network, and executing the segment routing or the traffic engineering path optimization to the bearing network to obtain the cross-layer cooperative terminal access bearing resource configuration; And executing the nearby user plane function selection or the session management strategy dynamic adjustment on the core network by means of the dynamic management strategy, and executing the video coding rate and resolution adjustment on the application server to obtain the cooperative adaptation parameters of the core network and the application layer.
9. The intelligent network quality of service management and control system based on real-time intent recognition and awareness as claimed in claim 1, wherein the scheduling manner of the cross-layer collaborative execution module includes: according to the matching analysis of the user movement track and the network coverage map in the intention prediction result, triggering double connection or conditional switching in advance for the service flow which is about to enter the weak coverage area, and obtaining resource reservation and switching preparation; And executing joint optimization configuration on the spectrum resources, the computing resources, the storage resources and the link resources through a resource orchestrator to obtain a resource configuration result of maximizing the resource utilization rate and guaranteeing the end-to-end service quality.

Description

Network service quality intelligent management and control system based on real-time intention recognition and perception Technical Field The invention discloses a network service quality intelligent management and control system based on real-time intention recognition and perception, and belongs to the technical field of digital information transmission. Background The intention is to drive the network as an important development direction in the field of network management, the core idea of which is to shift the center of gravity of network management from "how to configure" to "what result is desired". However, the existing network traffic quality control system still has a plurality of technical defects in terms of processing the fusion of the intention recognition and the network perception. Chinese patent CN110198237B discloses an "intention-driven network configuration method and apparatus", which parses an intention into a network configuration policy by acquiring a user intention description, and issues the policy to a network device for execution. The scheme solves the mapping problem of intent to configuration, but has the technical defects that firstly, the intent acquisition of the patent depends on explicit input of a user, implicit intent mining cannot be carried out based on real-time behavior and service flow characteristics of the user, intent dynamic change of the user in the using process is difficult to capture, secondly, the intent analysis of the patent is in a static conversion process, deep fusion with real-time network perceived data is lacking after the intent identification, and the intent understanding cannot be dynamically adjusted according to network state change, thirdly, the configuration strategy of the patent is generated based on a predefined rule base, deep reinforcement learning decision cannot be carried out for complex and changeable service flows, dynamic optimization is difficult to realize, and fourthly, the execution range of the patent is limited to network equipment configuration issuing, multilayer collaborative scheduling of a terminal, access, bearing, a core network and application cannot be realized, and an end-to-end service quality guarantee mechanism is lacking. In addition, the existing network sensing technology focuses on network state monitoring with single dimension, such as indexes of bandwidth utilization rate, time delay, packet loss rate and the like, and fails to perform associated modeling on multidimensional data such as user behavior, service characteristics, environment context and the like and user intention, so that a sensing result is disjointed from the real requirement of a user. Most of the existing control mechanisms are passive response modes of event triggering or periodic triggering, and lack of a dual-drive triggering mechanism based on intention change and state prediction makes active pre-adjustment difficult before service quality is reduced. Disclosure of Invention The invention aims to provide a network service quality intelligent management and control system based on real-time intention recognition and perception, which comprises a real-time intention recognition module, a multi-dimensional perception fusion module, an intelligent management and control decision module and a cross-layer collaborative execution module, wherein the real-time intention recognition module is used for carrying out implicit intention mining on real-time communication demands of a user according to user behavior data and service flow characteristics to obtain real-time user intention, the multi-dimensional perception fusion module is used for carrying out fusion modeling on multi-dimensional perception data by utilizing the real-time user intention and network state information to generate intention perception joint characteristics, the intelligent management and control decision module is used for carrying out deep reinforcement learning decision on network resources and service flows by virtue of the intention perception joint characteristics to obtain a dynamic management and control strategy, and the cross-layer collaborative execution module is used for completing cross-layer collaborative scheduling on terminal access bearing core network application by virtue of the dynamic management and control strategy to form end-to-end service quality guarantee. According to the method, the intention recognition is expanded from static analysis to real-time dynamic mining, the deep fusion of the intention and the perception is realized, the network management and control is upgraded from passive response to active prediction and self-adaptive optimization through intelligent decision and cross-layer cooperative execution driven by the intention, and the technical problems that the intention recognition and the perception are separated, the management and control response is lagged and difficult to cooperate end to end in the prior art are solved. Th