Design And Implementation Of Intent Automatic Translation Mechanism In Smart Integration Identifier Networking

With the rapid growth of the number of network users,complexity and types of network services,how to efficiently use and configure the network has become a hot issue.In recent years,with the maturity of artificial intelligence technology,the use of artificial intelligence technology to achieve network automation is a hot research direction.Among them,the Smart Integration Identifier Networking is a new network architecture proposed by Beijing Jiaotong University.This architecture uses the intent engine as the network brain,endows the network with intelligence,and conforms to the development trend of network automation.This paper designs and implements an intent automatic translation mechanism,which runs on the intent engine of the Smart Integration Identifier Networking,and realizes three important functions in the intent engine,namely:(1)Lightweight and flexible intent parsing;(2)Taking into account the effective mapping of physical resources and network resources;(3)Intention to issue,manage and repair.The main work and innovations of this paper are as follows:(1)This paper designs and implements a lightweight intent parsing method based on network knowledge graph,which realizes intent parsing from natural language.This method uses the network knowledge graph to pre-store the prior knowledge of the user’s network requirements,and analyzes the user’s requirements through natural language understanding.In addition,this method adds a secondary matching module to improve the accuracy of intent translation.Finally,the intent input in the form of natural language is transformed into the network policy and deployment policy in the form of descriptive language.(2)The method aims at minimizing delay and maximizing revenue,and considers the joint optimization of physical and computational resources.In addition,it also considers the location of the intent to access,and based on the intent mapping algorithm of two-stage mapping,it realizes the reasonable mapping of the parsing result on the physical network,taking into account the time complexity and effect of the algorithm.(3)This paper designs and implements an intent management and repair method to monitor and manage system resources and intents.This method designs multi-dimensional evaluation indicators to describe the operating status of the system,and realizes four management functions of intentions,namely,the issuance and registration of intentions,the uninstallation of expired intentions,the monitoring and repair of intentions,and the monitoring and management of system resources.Finally,in order to verify the function and effect of the automatic intent translation mechanism,this paper deploys the mechanism into the actual test system based on ONOS+Kubernetes,and conducts functional verification and performance testing.The final results show that the various methods in the automatic intent translation mechanism can work in coordination to achieve flexible and efficient intent parsing,mapping and distribution,and realize automatic fault repair and capacity expansion during intent running.In addition,compared with similar intent parsing and mapping methods,this mechanism ensures lower latency of intent translation,higher parsing accuracy,higher resource utilization and intent request acceptance rate.