CN114167760B - Intention driven network management system and method - Google Patents
Intention driven network management system and method Download PDFInfo
- Publication number
- CN114167760B CN114167760B CN202111236841.9A CN202111236841A CN114167760B CN 114167760 B CN114167760 B CN 114167760B CN 202111236841 A CN202111236841 A CN 202111236841A CN 114167760 B CN114167760 B CN 114167760B
- Authority
- CN
- China
- Prior art keywords
- network
- intention
- intent
- layer
- strategy
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000007726 management method Methods 0.000 claims abstract description 115
- 238000004458 analytical method Methods 0.000 claims abstract description 37
- 238000004891 communication Methods 0.000 claims abstract description 28
- 238000012795 verification Methods 0.000 claims abstract description 25
- 238000005516 engineering process Methods 0.000 claims abstract description 20
- 238000007418 data mining Methods 0.000 claims abstract description 13
- 239000010410 layer Substances 0.000 claims description 165
- 230000008859 change Effects 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 20
- 238000013468 resource allocation Methods 0.000 claims description 19
- 238000013473 artificial intelligence Methods 0.000 claims description 18
- 230000003993 interaction Effects 0.000 claims description 15
- 230000007246 mechanism Effects 0.000 claims description 15
- 238000003860 storage Methods 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 11
- 238000013461 design Methods 0.000 claims description 10
- 230000005540 biological transmission Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000009825 accumulation Methods 0.000 claims description 7
- 230000015654 memory Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 238000009826 distribution Methods 0.000 claims description 6
- 238000005111 flow chemistry technique Methods 0.000 claims description 6
- 238000009472 formulation Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 239000000203 mixture Substances 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 239000002346 layers by function Substances 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 230000006399 behavior Effects 0.000 claims description 4
- 238000011217 control strategy Methods 0.000 claims description 4
- 238000013500 data storage Methods 0.000 claims description 4
- 230000001788 irregular Effects 0.000 claims description 4
- 238000005065 mining Methods 0.000 claims description 4
- 230000007704 transition Effects 0.000 claims description 4
- ABEXEQSGABRUHS-UHFFFAOYSA-N 16-methylheptadecyl 16-methylheptadecanoate Chemical compound CC(C)CCCCCCCCCCCCCCCOC(=O)CCCCCCCCCCCCCCC(C)C ABEXEQSGABRUHS-UHFFFAOYSA-N 0.000 claims description 3
- 241000764238 Isis Species 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000005417 image-selected in vivo spectroscopy Methods 0.000 claims description 3
- 238000012739 integrated shape imaging system Methods 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 230000008447 perception Effects 0.000 description 11
- 238000011161 development Methods 0.000 description 8
- 230000018109 developmental process Effects 0.000 description 8
- 241000282414 Homo sapiens Species 0.000 description 6
- 230000009471 action Effects 0.000 description 5
- 238000013519 translation Methods 0.000 description 5
- 230000004927 fusion Effects 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 241001544487 Macromiidae Species 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013439 planning Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 102100035373 Cyclin-D-binding Myb-like transcription factor 1 Human genes 0.000 description 1
- 101000804518 Homo sapiens Cyclin-D-binding Myb-like transcription factor 1 Proteins 0.000 description 1
- 206010033799 Paralysis Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 239000002243 precursor Substances 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
- G05B19/0423—Input/output
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/25—Pc structure of the system
- G05B2219/25257—Microcontroller
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to the technical field of communication network management, and discloses an intention driving network management system and method, wherein after a user inputs a task, the user performs data mining analysis on the input task to obtain a user intention list; decoupling of a service application layer and a network control layer is realized through an intended northbound interface; the intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information; the southbound interface mainly adopts a virtualization technology to virtualize network element entities and network resources, and a more accurate implementation strategy output by an intention enabling layer is interacted with an infrastructure in a flow table mode; the state sensing module controls the underlying infrastructure sensors and actuators, and each sensor and actuator correspondingly modifies the configuration of the current infrastructure according to the received policy requirements, thereby realizing the requirements of users. The invention can reduce the reconfiguration of the network and improve the utilization rate of the network.
Description
Technical Field
The invention belongs to the technical field of communication network management, and particularly relates to an intention driving network management system and method.
Background
In recent years, network construction is used as a core of future informatization construction of China and is highly paid attention to by government of China. Network management is a basic activity of planning, organizing, supervising, controlling and billing by network operators and administrators for the use of network infrastructure and services, with the objective of ensuring stable, efficient and safe operation of the network. ISO (International Organization for Standardization ) proposes five functions of network management: fault management, configuration management, performance management, security management, and billing management. With the rapid development and widespread use of computer networks (particularly the Internet), the demands on the rapidity and effectiveness of network management have increased. Good network management system and technology are important foundation for keeping network safe, efficient and stable operation. With the continued development of networks, a variety of different network management models have emerged to manage network resources, such as SNMP (Simple Network Management Protocol ), DMI (Desktop Management Interface, desktop management interface), and the like.
The IETF proposes SNMP (Simple Network Management Protocol ), the precursor being SGMP (Simple Gateway Monitoring Protocol, simple gateway monitoring protocol). SGMP is mainly used for managing OSI three-layer routers, has simpler functions and cannot meet the demands of people. SGMP has further evolved and SNMP has evolved with the adoption of SMI (Structure of Management Information, management information structure) and MIB (Management Information Base ) architectures. SNMP mainly contains four operations: two for retrieving data (Get-Request and Get-Next-Request) and one for setting data (Set-Request) and issuing asynchronous notifications (Trap) to the device. SNMP is simple and easy to understand, easy to implement, occupies less network resources, is a management mode for performing fine configuration and hop-by-hop operation on equipment, and is relatively suitable for small-scale simple networks. As modern computer networks become huge in scale, network devices are large in variety and number and complex in relation, SNMP consumes more resources in a large-scale network, so that the balance between instantaneity and resource consumption cannot be realized; in addition, the MIB model adopted in SNMP is not suitable for storing a large amount of data and carrying out complex query operation; finally, SNMP cannot flexibly perform reconfiguration and modification of information, so that it is no longer suitable for the development of the current network.
Against the shortcomings of SNMP, the framework of PBNM (Policy-Based Network Management, policy-based management) proposed by IETF and DMTF. PBNM treats the network as a state machine, where policies are the basis for controlling and adjusting the state of the network. The policy mechanism enables an administrator to dynamically adjust the behavior of the whole system by only formulating new policies without changing management components in the system, and the network can automatically monitor and manage information access, network running states and network equipment according to the management policies formulated by the administrator, thereby automatically optimizing various operations required by the network. The PBNM framework is mainly composed of 4 functional components: (1) PMT (Policy Management Tool ): the PMT is an interface between the policy manager and the system, through which the network administrator edits, adds policies, and monitors the operational status of the overall policy control system. (2) PR (Policy Repository, policy knowledge base): PR is responsible for storing policy information of a policy system, and is typically implemented by using a directory server, or by using a relational database. If in the form of a directory server, the access protocol is a lightweight directory access protocol (Lightweight Directory Access Protocol, LDAP). (3) PDP (Policy Decision Point ): the PDP accepts the policy request of the policy execution point, makes policy decision on the request, acquires the corresponding policy from the policy knowledge base and transmits the policy to the policy execution point. (4) PEP (Policy Enforcement Point ): PEP is the final executor of the policy, and reports the network state change and policy execution condition.
With the current further expansion of network size and the continuous expansion of service dimension, policies in network management need to have higher level abstractions, IBNM (intelt-Based Network Management, intent-based network management) is proposed. Intent can be understood as a higher level policy, a rule expression after higher level abstraction of the policy; the IBNM can make a network administrator unnecessary to formulate a detailed policy, and does not need to know the overall condition of the network such as topology structure, link information and the like, and only the goal of expressing the service is needed to translate into a plurality of network policies for realizing the goal, and the network policies are converted into instructions for configuring and operating the equipment through the existing network management architecture. Along with the expansion of the scale and technical details of the network, the abstract degree can be improved to continuously shield more complex technical details, the network is managed by the intention of the highest level of abstraction, the mapping from the service intention to the system policy to the final detail configuration is formed, and the flexible management of the network is completed. A Network that uses intention to perform Network management is called an IDN (inter-drive Network).
SDN (Software Defined Network ) is an implementation of PBNM, and recent SDN developments began to support the implementation of IBNM. The SDN network architecture breaks through the limitation of network equipment, decouples a control layer and a data layer in the current network equipment, so that the network equipment becomes simple data forwarding equipment, and the network control layer is changed from a traditional logic centralized controller into a software-defined mode to realize a complex network control function. The typical SDN network architecture is divided into an application layer, a control layer and a data forwarding layer, wherein the control layer is interacted with the data forwarding layer through a southbound interface to acquire network state information and issue forwarding rules, and is interacted with the application layer through a northbound interface to realize business logic. Practice of SDN proves that obtaining low-level device information through a programming mode can lead to tight coupling between logic of network management and specific device characteristics, and the generality and portability are lacking; meanwhile, when the information of the network manager is queried, only a single device can be queried, and individual resources and group resources required by the application cannot be queried from the global angle. Based on the problems, the SDN needs to be further optimized and perfected, and the generation of IDN is promoted.
The network management DCN (Data Communication Network, integrated data communication network) is a communication system network between SDH (Synchronous Digital Hierarchy ) network elements, and communication transmission between SDH network elements and a management system. The DCN system can be implemented in various forms, and devices such as a BRIDGE (BRIDGE), a HUB (HUB/SWITCH), a ROUTER (ROUTER), a MODEM or an interface communication protocol converter are important contents for constructing a network management DCN management network. In the network management system, IP communication among NMS, EMS and access network element is mainly responsible for network management DCN; in addition, communication between various network elements is connected using a network management DCN system. The TCP/IP based connection path SMN communicates by means of a network management DCN. In transmission network management, because network management DCN monitoring is improper, some network elements are lost in transmission, so that the whole network is lost to be monitored, serious influence is generated, and network paralysis is caused; when the network transmission speed is low, the network delay is large, so that some application functions with strong functions are easy to cause and cannot be started to run; the problem of resetting the self-test by the network element device can lead to interruption of the transport network traffic. Based on the above factors, it is necessary to perform corresponding optimization on the network management DCN system, in order to ensure the smoothness of network element monitoring in the transmission network. Further pushing the birth of IDN.
At present, the rapid development of the mobile communication network is that the 5G is realized by 'everything interconnection', but the requirement of fusion application of the 5G and the vertical industry is still not clear, and the performance index of the 5G communication system can not meet the requirements of future wisdom city on landscape construction, the 6G network can effectively reduce cost and energy consumption, greatly improve the energy efficiency of the network and realize sustainable development. Compared with 5G, 6G has the characteristics of stronger performance, more intelligent, greener, wider coverage and the like. The 6G peak rate will reach 100 Gbit/s-1 Tbit/s; the air interface time delay is as low as 0.1ms; the connection number density supports 1000 ten thousand connections per square kilometer; the positioning accuracy will reach the order of centimeters. The 6G is more focused on the personalized needs of human beings as a center, and is continuously extended to space, overseas space, full-dimension perception world and network space, and a ubiquitous, non-time, non-human and non-human information infrastructure is provided for the human beings.
Future 6G networks need to achieve seamless fusion and perception of human-machine objects, abstraction and expression of five senses and intentions of people, and conversion from intentions to machine-recognizable operations. IDN (inter-Driven Network) is a Network service that can be automatically converted, verified, deployed, configured and optimized according to the user's intention to reach the target Network state, providing automation, high reliability and closed-loop optimization. Based on the result of user intention escape, the edge network needs to autonomously schedule the resources of wireless side network calculation, communication, storage and the like according to the state of the network, and autonomously select a proper artificial intelligent model for different management functions of configuration, optimization, failure and the like, and a dynamic decision deduction mechanism for resource fusion is provided, so that flexible arrangement and resource elastic utilization of the network are realized, the adaptation and management of network resources are continuously optimized, and the network service quality is continuously improved. Furthermore, by constructing a quality evaluation mechanism facing service characteristics, user characteristics and network environment, the method adapts to highly elastic change of the network, ensures self-evolution of cognitive process and learning, and enables the future 6G network to autonomously recognize network environment change and service characteristics.
Through the above analysis, the problems and defects existing in the prior art are as follows: the application of artificial intelligence in a network is limited to the optimization of a traditional network architecture, and the potential of the artificial intelligence is hardly fully exerted; the existing network management mainly comprises preset strategies, the management and control processes are relatively independent, and the problems of lack of cooperative optimization of a wireless side and a core side and the like are solved.
The difficulty of solving the problems and the defects is as follows: the artificial intelligence has some problems to be solved, including standardization of test cases and data sets for AI algorithm development and evaluation, influence of data integrity and accuracy on performance gain of the artificial intelligence algorithm, security reliability guarantee of the artificial intelligence in future communication, time delay control of the artificial intelligence algorithm and the like, which restrict the application of the artificial intelligence in network intelligent management and control.
The meaning of solving the problems and the defects is as follows: through the deep fusion with AI, the method is practically applied in multiple layers such as efficient transmission, seamless networking, endophytic safety, large-scale deployment, automatic maintenance and the like. In order to meet different service demands of high reliability, low time delay, high bandwidth and the like of the edge network users, the existing manual-based passive control mode is not applicable any more.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides an intention-driven network management system and method.
The present invention is thus embodied, an intention-driven network management method comprising the steps of:
step one, after a user inputs a task, further data mining analysis is carried out on the input task, and a user intention list is obtained after the task is fully constructed and understood, wherein the step can be used for fully interpreting the user intention;
and secondly, decoupling of a service application layer and a network control layer is realized through an intended northbound interface, network resources can be allocated according to requirements, and the difference of bottom network equipment is eliminated. The method comprises the steps of carrying out a first treatment on the surface of the
The intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information, and allows custom design intention to form an intention library, and the intention enabling layer can realize the mapping from intention to strategy;
step four, the southbound interface mainly adopts a virtualization technology to virtualize network element entities and network resources, and a more accurate implementation strategy which is output by an intention enabling layer is interacted with an infrastructure in a flow table mode;
And fifthly, the state sensing module controls the underlying infrastructure sensors and the actuators, and each sensor and each actuator correspondingly modify the configuration of the current infrastructure according to the received policy requirements, so that the requirements of users are realized.
Further, in the second step, the decoupling of the service application layer and the network control layer through the northbound interface includes:
as experience builds up, the authenticity of the intent generated is predicted, which if a given intent is complete and valid, is converted into a normalized, network-identifiable, intent tuple, defining intent tuple I D ＝<Domain, property, object, operation, result>Assume that the state space intended for mining is:
wherein s is ij Representing the jth value in the ith element in the intended five-tuple.
Further, in the second step, the decoupling of the service application layer and the network control layer through the northbound interface is further included:
named entity recognition based on BiLSTM-CRF named entity recognition model; extracting key entity information in irregular Chinese intentions through named entity identification, and outputting the parsed intentions; the model uses a method of combining a long-short-term memory neural network LSTM with a conditional random field CRF; translating and verifying the parsed intention, and outputting normalized intention tuples; the parsed intention is used for storing the form of the intention middle data structure; wherein the parser component marks the intent with the mark generator component while verifying it with the verifier component when parsing the intent.
Further, in the third step, the intention enabling layer performs four parts of intention text parsing, intention management, policy generation and policy verification on the received text information, and allows the custom design intention to form an intention library, including:
the intention text analysis is responsible for receiving the intention text translated by the intention northbound interface, extracting elements in the intention according to the attribute, wherein the elements comprise various performance requirements of the business; the intention management is responsible for controlling and managing the state transition of the intention; the strategy generation module is responsible for generating corresponding routing, security and resource allocation strategies according to the analyzed intention, and is realized by means of artificial intelligence and strategy library; policy verification is responsible for guaranteeing the completeness, correctness and feasibility verification of the intention of the policy; the network controller receives the strategy generated by the strategy decision module, combines the data information of the current network state, and adopts a learning training method to carry out comprehensive evaluation, analysis and judgment to complete flow processing among the cross-domain controllers, realize cooperative work among multiple controllers and further realize overall management optimization of the network; wherein the current network state includes network nodes, switches, servers, routes, communication resources, computing resources, and storage resources.
Further, the intention driven network management method further includes:
task-intent-policy loop: after the user task is input, the input task is subjected to further data mining analysis, and a user intention list is obtained after full deconstructing and understanding; the intent north interface translates intent input by a user, extracts main elements in the intent according to the attribute of the intent after translating the intent, such as the requirement of related business on performance in various aspects, and forms a flow related to the intent; processing the intention into a regularized intention request executed by the current network, and matching the intention after regularization with resources in a resource pool; generating corresponding routing, security and resource allocation strategies according to the analyzed intention, and realizing the routing, security and resource allocation strategies by means of artificial intelligence and strategy library; the generated strategy mechanism is used as a task to interact with the application layer through the northbound interface, and the feasibility of the generated decision is judged, so that a closed loop is formed; along with experience accumulation, through training of a large amount of data, the loop is enabled to automatically complete accurate analysis of intention, reasonable distribution of strategy and rational deployment of tasks, so that a programmable and customizable automatic management network is realized; the intelligent engine is used for completing the functions of data acquisition, data storage, data processing, model training and parameter adjustment, providing a certain priori condition for strategy formulation, and simultaneously ensuring the accuracy and reliability of the output network strategy and network configuration parameters through closed loop verification.
Perception-action loop: the loop utilizes interaction between a network environment sensor in the infrastructure and a controller in a network control layer, after the sensor in the network environment senses that network quality of service QoS and quantity information parameters of routing flows are changed, the sensor in the network environment obtains the changed information parameters to update the change and use conditions of each network element and various resources in the infrastructure in real time, the interacted information is stored in a network state sensing model, and is input to the network control layer, and the control layer forms a corresponding decision mechanism after receiving the state change, including how to change a topological structure and adjust routing weights, so as to drive the infrastructure to reconfigure and deploy autonomously; because the interaction between the network environment and the control layer is randomly changed in real time, the perception-action loop is expressed as a random Bayesian network which is unfolded according to time, and the performance of the whole network is reflected.
It is a further object of the present invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
After the user inputs the task, further data mining analysis is carried out on the input task, and a user intention list is obtained after the user is fully constructed and understood; decoupling of a service application layer and a network control layer is realized through an intended northbound interface; the intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information, and allows the custom design intention to form an intention library;
the southbound interface mainly adopts a virtualization technology to virtualize network element entities and network resources, and a more accurate implementation strategy output by an intention enabling layer is interacted with an infrastructure in a flow table mode; the state sensing module controls the underlying infrastructure sensors and actuators, and each sensor and actuator correspondingly modifies the configuration of the current infrastructure according to the received policy requirements, thereby realizing the requirements of users.
Another object of the present invention is to provide a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
after the user inputs the task, further data mining analysis is carried out on the input task, and a user intention list is obtained after the user is fully constructed and understood; decoupling of a service application layer and a network control layer is realized through an intended northbound interface; the intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information, and allows the custom design intention to form an intention library;
The southbound interface mainly adopts a virtualization technology to virtualize network element entities and network resources, and a more accurate implementation strategy output by an intention enabling layer is interacted with an infrastructure in a flow table mode; the state sensing module controls the underlying infrastructure sensors and actuators, and each sensor and actuator correspondingly modifies the configuration of the current infrastructure according to the received policy requirements, thereby realizing the requirements of users.
It is another object of the present invention to provide an intent-driven network management system including three functional hierarchies and three modules; the function layer comprises a business application layer, a network control layer and an infrastructure layer, and the modules comprise an intention analysis module, an intention enabling module and a state sensing module; the layers include two interfaces between layers: north and south interfaces are intended.
The functional layer comprises the following components from top to bottom:
and the service application layer is used for generating application intents by the user, namely, the user needs for different services or tasks under different scenes, and the network meets the user needs and further provides corresponding requirements for the network.
The network control layer has the functions of managing, controlling and making strategies for the network, collects network equipment information and overview network global, and executes cross-domain operation according to network resource distribution and link state information of all domains; the cross-domain operation comprises data transmission, load balancing and network resource allocation.
The state sensing layer is used for converting the physical resources and the functional resources of the bottom layer into abstract logic functional units based on the VNF technology, so as to realize resource pooling; the network state sensing layer is responsible for describing all available resource conditions in the current network; the network system is assumed to monitor the change of the state information of various resources in real time, update the resource change information table in real time, and store the change conditions of the computing resources, the communication resources and the storage resources in the state sensing layer, so as to provide data support for the intelligent network control layer.
The three modules included in the three layers are respectively:
the intention analysis module is used for converting the user intention into a network identifiable intention, constructing a mapping relation model between the user intention and the network identifiable intention by utilizing a related algorithm, establishing a connection relation, and ensuring that the network accurately identifies the user intention; the user inputs intention to analyze and verify links, if the given intention is complete and effective, the key entity information in the random Chinese intention can be extracted through named entity identification, and the analyzed intention is output; and translating and verifying the parsed intention, and outputting normalized intention tuples.
The intention enabling module is used for realizing management and control of the network, so as to complete related operations such as network resource management, network policy formulation and the like; after the controller receives the user intention transmitted by the intention northbound interface for the first time, the user intention is directly transmitted to a network state sensing module of the state sensing layer, and the state sensing module starts to collect related information of physical layers such as the current resource use condition of the underlying infrastructure, the running condition of equipment, the network congestion condition and the like after receiving the intention.
The state sensing module is used for interacting with the corresponding network environment, and comprises various network element devices in a space-based network, an access network and a ground network; updating the self parameters according to the interacted experience, storing the self parameters into a network state sensing model and simultaneously inputting the self parameters into a network control layer; various network nodes collect and transmit data to an intention engine and a controller of a control layer, and provide parameters for information feedback and strategy configuration; the adoption of network virtualization technology, namely redefining nodes in a traditional network, is a foundation for realizing the intention-driven network, provides a standardized control mechanism, and allows a control layer to control network behaviors and functions.
The system comprises two communication interfaces which are respectively:
the intent north interface is used for connecting the business application layer and the network control layer, and is used for analyzing and translating the intent of the user after the user intent is input by using natural language, translating the intent into a text which can be identified by a network consisting of an object, an operation and a result, namely, aiming at a certain network object, carrying out certain operation on the network object or expecting the object to present a certain result state; wherein the network object includes nodes, links, flows, and policies.
The southbound interface is used for connecting a network control layer and an infrastructure layer, takes a virtualization technology as a core, connects various network element devices, is used for interaction between the device layer and the control layer, virtualizes and slices various computing resources and communication resources, and improves the resource utilization rate of the network through flexible management; the southbound plug-ins fall into three main categories: topology collection OSPF, BGPLS, ISIS, configuration protocol NETCONF, SNMP, OVSDB, and guided by-pass Openflow, PCEP.
Further, the intent enabling module further includes:
the controller receives network situation sensing data which is transmitted by a northbound interface and fed back by a state sensing layer through a southbound interface, and then accurately and carefully learns and trains control strategy arrangement and deployment under various scenes, various resource scheduling and various sudden abnormal conditions, and a series of global parameters in a network are updated in real time according to learning results to allow custom design intents to form a meaning library in the management process of the network along with experience accumulation; generating corresponding routing, security and resource allocation strategies according to the analyzed intention, and realizing the routing, security and resource allocation strategies by means of artificial intelligence and strategy library; after the controllers receive the policy of decision, policy refinement and further configuration generation under SDN are realized, the controllers of different network domains are respectively deployed at different positions, the interfaces and communication protocols of the controllers have larger isomerism, and standard east/west interfaces can be independent of specific controllers to complete flow processing among cross-domain controllers, so that cooperative work among multiple controllers is realized, and further, integral management optimization of an integrated network is realized.
Another object of the present invention is to provide an information data processing terminal for implementing the intention-driven network management system.
By combining all the technical schemes, the invention has the advantages and positive effects that: the intention-driven network management system provided by the invention can be suitable for single application and combined application in various scenes, decouples a service application layer and a network control layer, can reduce network reconfiguration, improves the utilization rate of the network, eliminates human intervention in the network, and provides a good implementation framework for the intelligent development of a communication network.
Meanwhile, the invention can realize the intent input facing different scenes and different applications, flexibly and adaptively finish the reconfiguration of the network, ensure the normal realization of network demands, form an optimized closed-loop feedback loop and realize the automatic generation and the self-optimized allocation of the resource allocation strategy.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments of the present invention will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an intent-driven network management method according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of an intent-driven network management method according to an embodiment of the present invention.
Fig. 3 is a flowchart of an intent resolution module in an intent driven network management system according to an embodiment of the present invention.
Fig. 4 is a flowchart of an intent north interface in an intent-driven network management system according to an embodiment of the present invention.
Fig. 5 is a flowchart of an intention enabling module in an intention driven network management system according to an embodiment of the present invention.
Fig. 6 is a flowchart of a state sensing module in an intention driven network management system according to an embodiment of the present invention.
Fig. 7 is a task-intent-policy loop flowchart in an intent-driven network management technique provided by an embodiment of the present invention.
Fig. 8 is a flow chart of a sense-action loop in the intent-driven network management technique provided by an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The application scene of the embodiment is that in an integrated scene of the world, through interaction between a user and a machine (such as a mobile phone) on a ground station, intention is shown, deep mining and intention speculation of the user intention can be intelligently realized in an SAI (State-Action-Intent) architecture, a corresponding network strategy is formulated in a network, then the strategy is issued to an underlying facility (satellite) in a form of a flow table through a southbound interface, and the satellite starts to execute corresponding requirements after receiving a strategy task measurement instruction, so that user requirements are completed. The detailed description is as follows:
ground station human-machine interaction gives intent to: "I want to know that there are several cruisers in the Alaska sea area of the United states".
The intention enabling layer performs further text parsing according to the intention of the user: "America", "Alaska sea area", "several", "cruiser"; then generating a strategy: "schedule the in-orbit satellites 1, 2, 3 currently above the alaska sea area in the united states for data reconnaissance"; and after receiving the task information, the satellite feeds back the position information of the ground station, wherein the position information is respectively positioned at the current time, and the formulated strategy is further verified.
And (3) the three reconnaissance satellites of the infrastructure layer start radar remote sensing imaging on the current Alaska sea area, and after the number of cruisers is identified, the result is fed back to ground station personnel.
In view of the problems of the prior art, the present invention provides an intent-driven network management system and method, and the present invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the method for managing an intention-driven network according to the embodiment of the present invention includes the following steps:
s101, after a user inputs a task, further data mining analysis is carried out on the input task, and a user intention list is obtained after the user is fully constructed and understood;
s102, decoupling of a service application layer and a network control layer is achieved through an intended northbound interface;
s103, the intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information, and allows the custom design intention to form an intention library;
s104, the southbound interface mainly adopts a virtualization technology to virtualize network element entities and network resources, and the more accurate implementation strategy which is output by the intention enabling layer is interacted with the infrastructure in the form of a flow table;
s105, the state sensing module controls infrastructure sensors and actuators at the bottom layer, and each sensor and actuator correspondingly modifies the configuration of the current infrastructure according to the received policy requirements, so that the requirements of users are realized.
The schematic diagram of the method for managing the intention-driven network provided by the embodiment of the invention is shown in fig. 2.
The technical scheme of the invention is further described below with reference to specific embodiments.
Example 1
The intention driven network management system provided by the embodiment of the invention is mainly divided into three functional layers: a business application layer, a network control layer and an infrastructure layer; has three modules: intent resolution (Intent), network control (Action), and State awareness (State); there are two interfaces between layers: north and south interfaces are intended.
The three functional layers from top to bottom are respectively:
business application layer: the service application layer mainly generates application intention by a user, namely, the user needs different services or tasks under different scenes, and the network provides corresponding requirements for the network for meeting the user needs.
Network control layer: the network control layer has functions of managing, controlling and making strategies for the network, and can collect network equipment information, overview the network overall situation, and execute some cross-domain operations, such as data transmission, load balancing, network resource allocation and the like, according to the network resource distribution, link state and other information of all domains.
State perception layer: the network state sensing layer converts the physical resources and the functional resources of the bottom layer into abstract logic functional units based on VNF (Virtual Network Function) technology, and realizes resource pooling. The network state sensing layer is mainly responsible for describing all available resource conditions in the current network. The network system is assumed to monitor the change of the state information of various resources in real time, update the resource change information table in real time, and store the change conditions of the computing resources, the communication resources, the storage resources and the like in the state sensing layer, so that the data support is provided for the intelligent network control layer.
The three modules included in the three hierarchies are respectively:
intent resolution (intelt): the intention analysis module is a process of converting user intention into network identifiable intention, and utilizes a related algorithm to construct a mapping relation model between the user intention and the network identifiable intention, and establishes a connection relation to ensure that the network accurately identifies the user intention. And the user inputs the intention to analyze and verify links, if the given intention is complete and effective, the key entity information in the random Chinese intention can be extracted through named entity recognition, and the analyzed intention is output. And translating and verifying the parsed intention, and outputting normalized intention tuples.
Intent enabled (Action): the intention enabling module is used for realizing management and control of the network, thereby completing related operations such as network resource management, network policy making and the like. After the controller receives the user intention transmitted by the intention northbound interface for the first time, the user intention is directly transmitted to a network state sensing module of the state sensing layer, and the state sensing module starts to collect related information of physical layers such as the current resource use condition of the underlying infrastructure, the running condition of equipment, the network congestion condition and the like after receiving the intention. The controller can receive the data of the user intention transmitted by the northbound interface and the network situation perception fed back by the state perception layer through the southbound interface along with the accumulation of experience, and accurately and carefully learn and train the control strategy arrangement and deployment under various scenes, various resource scheduling, various sudden abnormal conditions and the like, and update a series of global parameters in the network in real time according to the learning result. In the process of managing the network, the custom design intent is allowed to form an intent library so as to improve the efficiency of network management based on the intent. Generating corresponding strategies such as routing, security, resource allocation and the like according to the analyzed intention, and realizing the strategies by means of artificial intelligence, strategy library and the like; after the controllers receive the policy of decision, policy refinement and further configuration generation under SDN are realized, the controllers of different network domains are respectively deployed at different positions, the interfaces and communication protocols of the controllers have larger isomerism, and standard east/west interfaces can be independent of specific controllers to complete flow processing among cross-domain controllers, so that cooperative work among multiple controllers is realized, and further, integral management optimization of an integrated network is realized.
State awareness (State): the state sensing module is interacted with the corresponding network environment and mainly comprises various network element devices in a space-based network, an access network and a ground network. And updating the parameters according to the interacted experience, storing the parameters into a network state sensing model and simultaneously inputting the parameters into a network control layer. Various network nodes collect and deliver data to the intention engine and controller of the control layer to provide parameters for information feedback and policy configuration. The adoption of network virtualization technology, namely redefining nodes in a traditional network, is a foundation for realizing the intention-driven network, provides a standardized control mechanism, and allows a control layer to control network behaviors and functions.
The system comprises two communication interfaces:
northbound interface (Intint NBI): the connection business application layer and the network control layer are used for translating the intention of the user into text which can be identified by a network consisting of an object, an operation and a result by analyzing and translating the intention of the user after the user is input in a natural language, namely, performing certain operation on a certain network object (including nodes, links, flows, strategies and the like) or expecting the object to present a certain result state.
Southbound interface (SBI): the network control layer and the infrastructure layer are connected, various network element devices are connected by taking a virtualization technology as a core, the network element devices are mainly used for interaction between the device layer and the control layer, various computing resources and communication resources are virtualized and sliced, and the resource utilization rate of the network is improved through flexible management. The southbound plug-ins are mainly divided into three main categories: topology collection (OSPF, BGPLS, ISIS), configuration protocol (NETCONF, SNMP, OVSDB), and guided by-pass (Openflow, PCEP).
The intention driving intelligent network management method provided by the embodiment of the invention comprises the following steps:
the first step: after the user inputs the task, further data mining analysis is carried out on the input task, and a user intention list is obtained after full deconstructing and understanding;
and a second step of: decoupling of the application layer and the network control layer is realized through the intention northbound interface, so that the diversity of application scenes, the diversity of intention main bodies and the diversity of intention demands can be met, the intention-driven network management system can form text formats (Json texts) which can be recognized by a network controller after intention abstraction after any main body (such as a person, a machine, an object and the like) inputs the intention into a network in any scene;
And a third step of: the intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information, allows custom design intention to form an intention library, and improves the efficiency of network management based on intention. The intention text analysis is responsible for receiving the intention text translated by the intention northbound interface, and extracting elements in the intention, such as various performance requirements of business, according to the attribute; the intention management is responsible for controlling and managing the state transition of the intention; the strategy generation module is responsible for generating corresponding strategies such as routing, security, resource allocation and the like according to the analyzed intention, and can be realized by means of artificial intelligence, strategy library and the like; policy verification is responsible for guaranteeing the completeness, correctness and feasibility verification of the intention of the policy. The network controller receives the strategy generated by the strategy decision module, combines the data information of the current network state (network node, switch, server, route, communication resource, computing resource, storage resource and the like), and adopts a learning and training method to comprehensively evaluate, analyze and judge, so as to complete flow processing among the cross-domain controllers, realize cooperative work among multiple controllers and further realize overall management optimization of the network;
Fourth step: the southbound interface mainly adopts a virtualization technology to virtualize network element entities and network resources, and a more accurate implementation strategy which is output by an intention enabling layer is interacted with an infrastructure in a flow table form so as to realize flexible management and improve the resource utilization rate of the network;
fifth step: the state sensing module controls the underlying infrastructure sensors and actuators, and each sensor and actuator correspondingly modifies the configuration of the current infrastructure according to the received policy requirements, so that the requirements of users can be realized.
The intention-driven intelligent network management technology provided by the embodiment of the invention mainly comprises the following two types:
task-intent-policy loop: after the user task is input, the input task is subjected to further data mining analysis, and a user intention list is obtained after full deconstructing and understanding; the intent north interface translates intent input by a user, extracts main elements in the intent according to the attribute of the intent after translating the intent, such as the requirement of related business on performance in various aspects, and forms a flow related to the intent; then, the intention is processed into a regularized intention request executed by the current network, and the intention after regularization is matched with resources in a resource pool; generating corresponding strategies such as routing, security, resource allocation and the like according to the analyzed intention, and realizing the strategies by means of artificial intelligence, strategy library and the like so as to ensure the completeness, correctness and feasibility verification of the intention by the strategy; and taking the generated strategy mechanism as a task, and carrying out interaction between the northbound interface and the application layer to judge the feasibility of the generated decision so as to form a closed loop. Along with experience accumulation, through training of a large amount of data, the loop can automatically complete accurate analysis of intention, reasonable distribution of strategies and rational deployment of tasks, so that a programmable and customizable automatic management network is realized. The intelligent engine is used for completing the functions of data acquisition, data storage, data processing, model training, parameter adjustment and the like, a certain priori condition is provided for strategy formulation, and meanwhile, the accuracy and reliability of the output network strategy and network configuration parameters are guaranteed through closed loop verification.
Perception-action loop: the loop utilizes interaction between a network environment sensor in the infrastructure and a controller in a network control layer, after the sensor in the network environment senses that information parameters such as network quality of service (QoS), the number of routing flows and the like are changed, the sensor in the network environment obtains the changed information parameters to update the change and use conditions of each network element and various resources in the infrastructure in real time, the interacted information is stored in a network state sensing model and is input to the network control layer, and the control layer forms a corresponding decision mechanism after receiving the state change, such as how to change a topological structure, adjust routing weights and the like, so as to drive the infrastructure to reconfigure and deploy autonomously. Because the interaction between the network environment and the control layer is randomly changed in real time, the perception-action loop can be expressed as a random Bayesian network which is unfolded according to time, and the performance of the whole network can be reflected. The formation of the self-optimizing perception-action closed loop can greatly improve the overall perception and control capability of the network state.
Example 2
The intention driving network management method provided by the embodiment of the invention comprises the following steps:
Step one, after a user inputs a task, further data mining analysis is carried out on the input task, and a user intention list is obtained after the user is fully constructed and understood;
step two, the Northbound interface (Intint NBI) implementation is intendedDecoupling of the business application layer and the network control layer, along with the accumulation of experience, predicts the authenticity of the generated intent, converts a given intent, if complete and valid, into normalized, network-identifiable intention tuples, defining an intention quintuple I D ＝<Domain, property, object, operation, result>Assume that the state space intended for mining is:
wherein s is ij Representing the jth value, e.g. s, in the ith element in the intended quintuple 13 Representing the third attribute in the set of domain attributes in the intent quintuple. And then extracting key entity information in the irregular Chinese intention through named entity identification, and outputting the analyzed intention. The named entity recognition is mainly an entity extraction algorithm based on a named entity recognition model of BiLSTM-CRF. The model uses a Long short term memory neural network (Long Shot-Term Memory Neural Network, LSTM) combined with a conditional random field (Conditional Random Field, CRF). And translating and verifying the parsed intention, and outputting normalized intention tuples. The parsed intention is used for storing the form of the intention middle data structure. Wherein the parser component marks the intent with the mark generator component while verifying it with the verifier component when parsing the intent. The normalized network intention is input to assist the network intention to improve the speed and the accuracy of network decision, and interactive iterative updating is continuously carried out with the network intention.
And thirdly, after the network control layer receives the user intention transmitted through the intention northbound interface for the first time, the user intention is directly transmitted to a network state sensing module of the state sensing layer, and the state sensing module starts to collect related information of the physical layers such as the current resource use condition of the underlying infrastructure, the running condition of equipment, the network congestion condition and the like after receiving the intention. The controller receives data of user intention transmitted by the northbound interface, network situation perception fed back by the state perception layer through the southbound interface, and the like, and accurately and carefully learns and trains control strategy arrangement and deployment under various scenes, various resource scheduling, various sudden abnormality and the like, and updates a series of global parameters in the network in real time according to learning results.
Step four, the southbound interface SBI (Southbound Interface) realizes decoupling of a control plane and a data plane of the network, so that a control platform can provide lower-layer data abstraction, and application programming is simplified; meanwhile, the controller of the network has network global information, so that more consistent and effective decisions can be made.
And fifthly, the state sensing module executes a series of intention tasks issued by the network control module, interacts with the corresponding network environment according to the requirements, updates own parameters according to the interacted experience, stores the parameters into a network state sensing model, and transmits the parameters to a network control layer through a southbound interface to assist in the formulation of strategies.
Step six, after the task is issued by the user (man, machine, object), the task needs to be subjected to intention analysis and intention translation into text which can be read and understood by a machine (such as Json text), the translated intention is input into a network control layer for strategy arrangement and deployment, and interaction is performed between the user and an upper application layer to form a closed loop.
Step seven, the bottom-up perception-action loop represents the interaction between the network environment and the infrastructures, and each infrastructure deployed at the bottom layer has a corresponding learning mechanism and can perform interactive learning among the infrastructures. Wherein the agent is intended to influence the network environmental state through the network and to receive a perception of the environmental state through its sensor. The agent selects the action of the next step according to the perception information of the current time step. The dynamics affect the state of the environment and thus the input of the smart in the next time step, and the cycle is repeated.
The application principle of each module of the present invention is described in further detail below with reference to the accompanying drawings.
The application scene of the invention is a process which can be compatible with and adaptive to various application scenes, and mainly considers the automatic driving management control of the network in the network, and mainly comprises three modules: an intention analysis module (Intent), an intention enabling module (Action) and a State sensing module (State); two interfaces: the northbound interface (intelntnbi) is intended to implement decoupling of the business application layer from the network control plane, and the southbound interface (SBI) is intended to implement decoupling between the data business layer, i.e., the network state aware layer, and the network control layer.
1) Intention analysis module (Intent)
As shown in fig. 3, the main flow is described in detail below.
Step one: the user (people, machines and objects) issues a task command according to the requirements of the application scene; if the task command is accurate, the task command can be directly input; if the command of the task command is not clear and accurate and the coverage range is wide, the task needs to be further analyzed, excavated and tunneled;
step two: deep data mining is carried out on the task, and the task is fully deconstructed and understood;
step three: a detailed list of user intents is output.
2) North interface intention (IntntNBI)
As shown in fig. 4, the detailed flow of the northbound interface is resolved as follows:
step one: inputting a user intention list;
step two: intention verification
Analyzing and verifying the authenticity of the generated intention, constructing a mapping relation model between the user intention and the network identifiable intention by using a related algorithm, establishing a connection relation, ensuring that the network accurately identifies the user intention so as to assist the user intention to improve the decision speed and accuracy;
step three: intent translation
Intent translation allows a user to input declarative intent in different forms, such as graphics, speech, text, etc., and the user can express intent in natural language or directly describe parameter requirements related to network functions. If the given intention is complete and effective, extracting key entity information in the irregular Chinese intention through named entity identification, and outputting the analyzed intention. The user's intention in the application layer expressed in approximately "natural language" is translated into a network intention consisting of "objects", "operations" and "results", i.e. for a certain network object (including nodes, links, flows and policies, etc.), it is subjected to a certain kind of operation, or the object is expected to present a certain result state. And converting the user intention into a language recognized by the network based on the morphological rule of the intention quintuple by using NLP and AI auxiliary analysis technology, and taking the output intention quintuple as the input of the intelligent network management and control requirement.
Step four: intention verification
And adopting a consistency check framework based on a commitment theory to carry out overall planning on people, machines and objects, taking an infrastructure module as an intention executor, providing commitments of the bottom network capability set to an upper layer, taking an intention understanding module as a policy maker, and receiving the commitments from the infrastructure module. And finishing verification of whether the translation result can completely match the original intention of the user or not, and feeding back the verification result to the user.
3) Intention enabling module (Action)
As shown in fig. 5, the main flow is described in detail.
Step one: inputting text which can be recognized by a machine;
step two: the intention text analysis is responsible for receiving the intention text translated by the intention northbound interface, and extracting elements in the intention, such as various performance requirements of business, according to the attribute;
step three: intent management, which is responsible for controlling and managing the state transition of intent;
step four: the strategy generation is responsible for generating corresponding strategies such as routing, security, resource allocation and the like according to the analyzed intention, and can be realized by means of artificial intelligence, strategy library and the like;
step five: and the policy verification is responsible for guaranteeing the completeness, correctness and feasibility verification of the intention of the policy.
4) Southward pointing interface (SBI)
The network element entity and the network resource are virtualized mainly by adopting a virtualization technology, and interaction is carried out between the network element entity and the infrastructure in the form of a flow table;
5) State perception module (State)
As shown in fig. 6, the main flow is described in detail.
Step one: collecting relevant information of each device at the bottom layer and environmental state change, and storing the collected information in a state sensing module;
step two: the stored information is subjected to coarse granularity division by adopting a Classification (Classification) method in machine learning, and the divided state information is described in a numbering mode;
step three: carrying out ontology abstraction on each infrastructure and resources of the bottom layer with finished numbers, and specifically defining the following in the form of concepts, relations, attributes and constraint quadruples:
O:＝{C,R,Att,F}，
wherein C represents a set of all concepts in the network resource ontology; r represents a set of all relationships in the network ontology definition, and specifically describes the entity in the network resource ontology and the set of the relationships between the entity and the attribute; att represents a set of network resource attributes; f represents the constraint of the attribute.
Step four: and transmitting the processed information of the state sensing layer to a network control layer controller through a southbound interface.
Two aspects of the present invention are described in further detail below with reference to the accompanying drawings.
As shown in fig. 7, the task-intent-policy loop detailed flow is described in detail:
step one: user task input, further data mining and analysis are carried out on the input task, and a user intention list is obtained after full deconstructing and understanding;
step two: performing intent translation and verification on the user intent, and outputting a regularized intent request;
step three: analyzing the intent text after regularization, and generating corresponding strategies such as routing, security, resource allocation and the like through means such as artificial intelligence, strategy library and the like;
step four: the generated strategy mechanism is used as a task to interact with the application layer through the northbound interface, and the feasibility of the generated decision is judged, so that a closed loop is formed.
As shown in fig. 8, the sense-action loop detailed flow is described in detail:
step one: after the sensor of the equipment in the network environment senses the change of information parameters such as network quality of service (QoS), the number of routing flows and the like;
step two: updating the changed information parameters to parameter tables of various network elements and various resources in the infrastructure in real time, storing the interacted information into a network state sensing model, and transmitting the interacted information to an upper network control layer;
Step three: after receiving the state change, the control layer forms a corresponding decision mechanism, such as how to change the topology structure, adjust the routing weight, and the like, and then interacts with the infrastructure layer to drive the infrastructure to reconfigure and deploy autonomously, thereby forming a closed loop.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When used in whole or in part, is implemented in the form of a computer program product comprising one or more computer instructions. When loaded or executed on a computer, produces a flow or function in accordance with embodiments of the present invention, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL), or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk SolidStateDisk (SSD)), etc.
The foregoing is merely illustrative of specific embodiments of the present invention, and the scope of the invention is not limited thereto, but any modifications, equivalents, improvements and alternatives falling within the spirit and principles of the present invention will be apparent to those skilled in the art within the scope of the present invention.
Claims (8)
1. An intent-driven network management system, comprising three functional layers and three modules; the function layer comprises a service application layer, a network control layer and a state sensing layer, and the modules comprise an intention analysis module, an intention enabling module and a state sensing module; the layers include two interfaces: the intent north interface and south interface;
the functional layer comprises the following components from top to bottom:
the business application layer is used for generating application intents by users, namely, the demands of the users on different businesses or tasks under different scenes, and the network meets the demands of the users and further provides corresponding demands for the network;
the network control layer has the functions of managing, controlling and making strategies for the network, collects network equipment information and overview network global, and executes cross-domain operation according to network resource distribution and link state information of all domains; the cross-domain operation comprises data transmission, load balancing and network resource allocation;
The state sensing layer is used for converting the physical resources and the functional resources of the bottom layer into abstract logic functional units based on the VNF technology, so as to realize resource pooling; the state sensing layer is responsible for describing all available resource conditions in the current network; the network system is assumed to monitor the change of the resource state information in real time, update the resource change information table in real time, and store the change conditions of the computing resource, the communication resource and the storage resource in the state sensing layer so as to provide data support for the intelligent network control layer;
the three modules included in the three hierarchies are respectively:
the intention analysis module is used for converting the user intention into a network identifiable intention, constructing a mapping relation model between the user intention and the network identifiable intention by utilizing a related algorithm, establishing a connection relation, and ensuring that the network accurately identifies the user intention; the user inputs intention to analyze and verify links, if the given intention is complete and effective, the key entity information in the random Chinese intention can be extracted through named entity identification, and the analyzed intention is output; translating and verifying the parsed intention, and outputting normalized intention tuples;
the intention enabling module is used for realizing management and control of the network so as to finish related operations of network resource management and network policy formulation; the controller directly transmits the user intention transmitted through the intention northbound interface to the network state sensing module of the state sensing layer after receiving the intention for the first time, and the state sensing module starts to collect the current resource use condition of the underlying infrastructure, the running condition of the equipment and the related information of the physical layer of the network congestion condition after receiving the intention;
The state sensing module is used for interacting with the corresponding network environment and comprises a space-based network, an access network and network element equipment in a ground network; updating the self parameters according to the interacted experience, storing the self parameters into a network state sensing model and simultaneously inputting the self parameters into a network control layer; the network node collects and transmits the data to an intention engine and a controller of a control layer, and provides parameters for information feedback and strategy configuration; the network virtualization technology is adopted, namely redefining of nodes in the traditional network is a foundation for realizing the intention-driven network, a standardized control mechanism is provided, and a control layer is allowed to control network behaviors and functions;
the system comprises two communication interfaces which are respectively:
the intent north interface is used for connecting the business application layer and the network control layer, and is used for analyzing and translating the intent of the user after the user intent is input by using natural language, translating the intent into a text which can be identified by a network consisting of an object, an operation and a result, namely, aiming at a certain network object, carrying out certain operation on the network object or expecting the object to present a certain result state; wherein the network object includes nodes, links, flows, and policies;
The southbound interface is used for connecting a network control layer and a state sensing layer, taking a virtualization technology as a core, connecting network element equipment, being used for interaction between the equipment layer and the control layer, carrying out virtualization and slicing on computing resources and communication resources, and improving the resource utilization rate of a network through flexible management; the southbound plug-ins fall into three main categories: topology collection OSPF, BGPLS, ISIS, configuration protocol NETCONF, SNMP, OVSDB and guided by-pass Openflow, PCEP;
the intent enabled module further includes:
the controller receives network situation sensing data which is transmitted by a northbound interface and fed back by a state sensing layer through a southbound interface, and then accurately and carefully learns and trains control strategy arrangement and deployment under various scenes, various resource scheduling and various sudden abnormal conditions, and a series of global parameters in a network are updated in real time according to learning results to allow custom design intents to form a meaning library in the management process of the network along with experience accumulation; generating corresponding routing, security and resource allocation strategies according to the analyzed intention, and realizing the routing, security and resource allocation strategies by means of artificial intelligence and strategy library; after the controllers receive the policy of decision, policy refinement and further configuration generation under SDN are realized, the controllers of different network domains are respectively deployed at different positions, the interfaces and communication protocols of the controllers have isomerism, and standard east/west interfaces can be independent of specific controllers to complete flow processing among cross-domain controllers, so that cooperative work among multiple controllers is realized, and overall management optimization of an integrated network is further realized.
2. An intention-driven network management method applied to the intention-driven network management system of claim 1, characterized in that the intention-driven network management method comprises the steps of:
step one, after a user inputs a task, further data mining analysis is carried out on the input task, and a user intention list is obtained after the user is fully constructed and understood;
decoupling a service application layer and a network control layer through an intended northbound interface;
the intention enabling layer carries out intention text analysis, intention management, strategy generation and strategy verification on the received text information, and allows the custom design intention to form an intention library, and the method specifically comprises the following steps:
the intention text analysis is responsible for receiving the intention text translated by the intention northbound interface, extracting elements in the intention according to the attribute, wherein the elements comprise performance requirements of the business; the intention management is responsible for controlling and managing the state transition of the intention; the strategy generation module is responsible for generating corresponding routing, security and resource allocation strategies according to the analyzed intention, and is realized by means of artificial intelligence and strategy library; policy verification is responsible for guaranteeing the completeness, correctness and feasibility verification of the intention of the policy; the network controller receives the strategy generated by the strategy decision module, combines the data information of the current network state, and adopts a learning training method to carry out comprehensive evaluation, analysis and judgment to complete flow processing among the cross-domain controllers, realize cooperative work among multiple controllers and further realize overall management optimization of the network; wherein the current network state includes network nodes, switches, servers, routes, communication resources, computing resources, and storage resources;
Step four, the southbound interface adopts a virtualization technology to virtualize network element entities and network resources, and a more accurate implementation strategy output by an intention enabling layer is interacted with an infrastructure in a flow table mode;
and fifthly, the state sensing module controls the underlying infrastructure sensors and the actuators, and each sensor and each actuator correspondingly modify the configuration of the current infrastructure according to the received policy requirements, so that the requirements of users are realized.
3. The method for managing an intended drive network as set forth in claim 2, wherein in the second step, the decoupling of the service application layer and the network control layer through the intended northbound interface includes:
as experience builds up, the authenticity of the intent generated is predicted, and if a given intent is complete and valid, it is converted into a normalized, network-identifiable, intent tuple, defining the intent tupleAssume that the state space intended for mining is:
；
wherein,representing the jth value in the ith element in the intended five-tuple.
4. The method for managing an intended drive network as recited in claim 2, wherein in the second step, the decoupling of the service application layer and the network control layer is implemented through an intended northbound interface, further comprising:
Named entity recognition based on BiLSTM-CRF named entity recognition model; extracting key entity information in irregular Chinese intentions through named entity identification, and outputting the parsed intentions; the model uses a method of combining a long-short-term memory neural network LSTM with a conditional random field CRF; translating and verifying the parsed intention, and outputting normalized intention tuples; the parsed intention is used for storing the form of the intention middle data structure; wherein the parser component marks the intent with the mark generator component while verifying it with the verifier component when parsing the intent.
5. The intent-driven network management method as recited in claim 2, wherein the intent-driven network management method further comprises:
task-intent-policy loop: after the user task is input, the input task is subjected to further data mining analysis, and a user intention list is obtained after full deconstructing and understanding; the intent north interface translates intent input by a user, extracts elements in the intent according to the attribute of the intent after translating the intent, including the requirement of related business on performance, and forms a flow related to the intent; processing the intention into a regularized intention request executed by the current network, and matching the intention after regularization with resources in a resource pool; generating corresponding routing, security and resource allocation strategies according to the analyzed intention, and realizing the routing, security and resource allocation strategies by means of artificial intelligence and strategy library; the generated strategy mechanism is used as a task to interact with the application layer through the northbound interface, and the feasibility of the generated decision is judged, so that a closed loop is formed; along with experience accumulation, through training of a large amount of data, the loop is enabled to automatically complete accurate analysis of intention, reasonable distribution of strategy and rational deployment of tasks, so that a programmable and customizable automatic management network is realized; the intelligent engine is used for completing the functions of data acquisition, data storage, data processing, model training and parameter adjustment, providing a certain priori condition for strategy formulation, and simultaneously performing closed loop verification to ensure the accuracy and reliability of the output network strategy and network configuration parameters;
Perception-action loop: the loop utilizes interaction between a network environment sensor in the infrastructure and a controller in a network control layer, after the sensor in the network environment senses that network quality of service QoS and quantity information parameters of routing flows are changed, the sensor in the network environment obtains the changed information parameters to update the change and use conditions of each network element and various resources in the infrastructure in real time, the interacted information is stored in a network state sensing model, and is input to the network control layer, and the control layer forms a corresponding decision mechanism after receiving the state change, including how to change a topological structure and adjust routing weights, so as to drive the infrastructure to reconfigure and deploy autonomously; because the interaction between the network environment and the control layer is randomly changed in real time, the perception-action loop is expressed as a random Bayesian network which is unfolded according to time, and the performance of the whole network is reflected.
6. A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the method steps of claim 2.
7. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the method steps of claim 2.
8. An information data processing terminal for implementing the intention-driven network management system according to claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111236841.9A CN114167760B (en) | 2021-10-23 | 2021-10-23 | Intention driven network management system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111236841.9A CN114167760B (en) | 2021-10-23 | 2021-10-23 | Intention driven network management system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114167760A CN114167760A (en) | 2022-03-11 |
| CN114167760B true CN114167760B (en) | 2023-12-29 |
Family
ID=80477240
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111236841.9A Active CN114167760B (en) | 2021-10-23 | 2021-10-23 | Intention driven network management system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114167760B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114697219B (en) * | 2022-03-24 | 2024-03-08 | 阿里巴巴（中国）有限公司 | Network control method, equipment and system for live network |
| CN115065596B (en) * | 2022-03-29 | 2023-09-26 | 上海交通大学 | Industrial heterogeneous network integrated configuration system and method based on software definition |
| CN117062113A (en) * | 2022-05-07 | 2023-11-14 | 中国移动通信有限公司研究院 | Method, device, network equipment and storage medium for configuring slicing strategy |
| CN117376264A (en) * | 2022-06-30 | 2024-01-09 | 华为技术有限公司 | Information acquisition method, device, system, storage medium and computer program product |
| CN117750407A (en) * | 2022-09-14 | 2024-03-22 | 中兴通讯股份有限公司 | Intent-based telecommunication network management method and device |
| CN115412369B (en) * | 2022-10-31 | 2022-12-27 | 中国人民解放军军事科学院系统工程研究院 | Method and system for constructing toughness atom service framework facing network control |
| CN115695135B (en) * | 2022-12-30 | 2023-03-21 | 鹏城实验室 | Active network management method, device, equipment and storage medium |
| CN116137609B (en) * | 2023-02-27 | 2024-04-16 | 西安电子科技大学 | Hierarchical routing method for intention-driven cross-domain data link network |
| CN115994541B (en) * | 2023-03-22 | 2023-07-07 | 金蝶软件（中国）有限公司 | Interface semantic data generation method, device, computer equipment and storage medium |
| CN116107573B (en) * | 2023-04-12 | 2023-06-30 | 广东省新一代通信与网络创新研究院 | Intention analysis method and system based on finite state machine |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9787639B1 (en) * | 2016-06-24 | 2017-10-10 | Varmour Networks, Inc. | Granular segmentation using events |
| CN109245916A (en) * | 2018-08-15 | 2019-01-18 | 西安电子科技大学 | A kind of the cloud access net system and method for intention driving |
| CN110278111A (en) * | 2019-05-29 | 2019-09-24 | 西安电子科技大学 | A kind of intention driving network generic structure and its it is intended to driving network translation method |
| CN112565336A (en) * | 2020-11-06 | 2021-03-26 | 西安电子科技大学 | Intelligent Internet of things centralized control method, system, medium, equipment and application |
| CN112565193A (en) * | 2020-11-06 | 2021-03-26 | 西安电子科技大学 | Network security policy conflict resolution method, system, storage medium and equipment |
| CN112631856A (en) * | 2020-12-17 | 2021-04-09 | 西安电子科技大学 | Intelligent operation and maintenance method, system, equipment and application of intention-driven 6G network |
| CN112636958A (en) * | 2020-12-12 | 2021-04-09 | 东北大学 | Policy autoverification techniques in intent-driven networks |
| CN112953778A (en) * | 2021-03-17 | 2021-06-11 | 北京交通大学 | Intention-driven-based service arrangement system and method in intelligent fusion identification network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102105683B1 (en) * | 2017-04-28 | 2020-05-29 | 한국전자통신연구원 | Integrated Platform Management Device And Method For Wire and Mobile communication Service |
-
2021
- 2021-10-23 CN CN202111236841.9A patent/CN114167760B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9787639B1 (en) * | 2016-06-24 | 2017-10-10 | Varmour Networks, Inc. | Granular segmentation using events |
| CN109245916A (en) * | 2018-08-15 | 2019-01-18 | 西安电子科技大学 | A kind of the cloud access net system and method for intention driving |
| CN110278111A (en) * | 2019-05-29 | 2019-09-24 | 西安电子科技大学 | A kind of intention driving network generic structure and its it is intended to driving network translation method |
| CN112565336A (en) * | 2020-11-06 | 2021-03-26 | 西安电子科技大学 | Intelligent Internet of things centralized control method, system, medium, equipment and application |
| CN112565193A (en) * | 2020-11-06 | 2021-03-26 | 西安电子科技大学 | Network security policy conflict resolution method, system, storage medium and equipment |
| CN112636958A (en) * | 2020-12-12 | 2021-04-09 | 东北大学 | Policy autoverification techniques in intent-driven networks |
| CN112631856A (en) * | 2020-12-17 | 2021-04-09 | 西安电子科技大学 | Intelligent operation and maintenance method, system, equipment and application of intention-driven 6G network |
| CN112953778A (en) * | 2021-03-17 | 2021-06-11 | 北京交通大学 | Intention-driven-based service arrangement system and method in intelligent fusion identification network |
Non-Patent Citations (2)
| Title |
|---|
| 夏洋 ; 黄治华 ; .基于NFV和SDN的通信资源虚拟化及调度研究.信息与电脑(理论版).2018,(17),全文. * |
| 魏翼飞 ; 李莉 ; .面向能效的可编程网络关键技术探讨.移动通信.2016,(10),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114167760A (en) | 2022-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114167760B (en) | Intention driven network management system and method | |
| CN110278111B (en) | General architecture of intention-driven network and translation method of intention-driven network | |
| CN109245916B (en) | Intention-driven cloud access network system and method | |
| CN105976031B (en) | Parallel processing of data by multiple semantic inference engines | |
| CN108259367B (en) | Service-aware flow strategy customization method based on software defined network | |
| CN112600717B (en) | Satellite network management and control protocol semi-physical test device based on SDN | |
| Yang et al. | Intent defined optical network with artificial intelligence-based automated operation and maintenance | |
| Carrera et al. | A real-life application of multi-agent systems for fault diagnosis in the provision of an Internet business service | |
| Khowaja et al. | Toward energy-efficient distributed federated learning for 6G networks | |
| CN110213175A (en) | A kind of intelligent managing and control system and management-control method towards knowledge definition network | |
| Yang et al. | Smart intent-driven network management | |
| CN116170314B (en) | Digital twin network platform device for AI (advanced technology attachment) endophytic under network slice scene | |
| CN110309318A (en) | The intention of communication network characterizes system and method, information data processing terminal | |
| CN112543170A (en) | Software-defined wide area network management method, device and system, and storage medium | |
| WO2022083549A1 (en) | Traffic signal conversion method and apparatus, electronic device, and storage medium | |
| Rezazadeh et al. | SliceOps: Explainable MLOps for Streamlined Automation-Native 6G Networks | |
| Huang et al. | A brief survey and implementation on ai for intent-driven network | |
| CN102045186B (en) | Event analysis method and system | |
| CN117640335A (en) | Dynamic adjustment and optimization method for intelligent building comprehensive wiring | |
| Tong et al. | Ten issues of NetGPT | |
| CN116170878A (en) | Intent conflict coordination solution method based on improved IBN network | |
| CN112819380B (en) | Power grid regulation and control panoramic transparent access method, system, equipment and storage medium | |
| CN115053188A (en) | Event prediction | |
| Wang et al. | Slice management mechanism based on dynamic weights for service guarantees in smart grid | |
| Nsaif et al. | SM-FPLF: Link-State Prediction For Software-Defined DCN Power Optimization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |