CN111683384A - Network optimization method for realizing dynamic weighting of communication link by artificial intelligence - Google Patents
Network optimization method for realizing dynamic weighting of communication link by artificial intelligence Download PDFInfo
- Publication number
- CN111683384A CN111683384A CN202010521746.2A CN202010521746A CN111683384A CN 111683384 A CN111683384 A CN 111683384A CN 202010521746 A CN202010521746 A CN 202010521746A CN 111683384 A CN111683384 A CN 111683384A
- Authority
- CN
- China
- Prior art keywords
- communication
- link
- path
- network
- jump
- 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.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/22—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks comprising specially adapted graphical user interfaces [GUI]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/24—Connectivity information management, e.g. connectivity discovery or connectivity update
- H04W40/248—Connectivity information update
Abstract
The invention provides a network optimization method for realizing dynamic weighting of a communication link by using artificial intelligence, which relates to the technical field of dynamic weighting optimization of the communication link and comprises the following steps: the communication element acquires all link characteristics of a communication path of the communication element, records the link characteristics in real time, forms a link table, and generates an intelligent visual network topological graph according to the link characteristics of the link table; when a communication request exists, identifying an intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics among communication elements, importing the link characteristics into a link image identification sample library for comparison, calculating a real-time dynamic weighted value of each link, calculating an optimal communication path, and realizing communication according to the optimal communication path; and updating the intelligent visual network topological graph according to the latest link characteristics participating in the communication, introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning, forming a latest link image recognition sample library, and finishing optimization.
Description
Technical Field
The invention relates to the technical field of dynamic weighting optimization of communication network links, in particular to a network optimization method for realizing dynamic weighting of communication links by using artificial intelligence.
Background
At present, fixed protocols such as IEEE802.11 series and internet of things 802.15.4 are basically adopted in networking communication to realize networking communication of a local area network or a small network. In consideration of empirical calculation values such as network reliability and robustness, the networking process usually needs to perform modes such as multiple broadcast routing requests, communication handshake and time slot control, and then determines an effective link according to a fixed algorithm weight. There are mainly the following problems:
1) in a high-speed network, 4G, 5G, WIFI, etc. delays and collisions are not significant, but network redundancy, congestion, also occurs as a result;
2) in a low-speed network, such as zigbee, lora, aloha and other internet of things, network collision and delay caused by information concurrency are particularly obvious.
Moreover, the optimization method of the wireless local area network and the small ad hoc network depends on the consideration of empirical calculation values such as network reliability and robustness, and the networking process usually needs to be realized by performing modes such as multiple broadcast routing requests, communication handshake, time slot control and the like, and then determining an effective link according to the weighting of a fixed algorithm. There are mainly the following problems:
1) in high-speed internet and internet of things networks, such as protocols based on IEEE802.11 such as OFDM and WIFI, delay collision is not obvious, but equipment in the group network and channel state real-time monitoring are lacked, network redundancy congestion is caused by multiple routing requests and handshake protocols, once the network has node performance reduction and the like, other optimal paths cannot be quickly searched, the network performance can be quickly reduced, and the network performance is ensured by adopting a method of regularly or manually cleaning channels. And the fixed weighting algorithm can not realize effective dynamic adjustment according to the complex environment change.
2) In low-speed internet of things, such as networks based on IEEE802.15.4 or using a custom communication protocol, e.g. zigbee, lora, aloha, etc., most of network performances adopt a time-sharing multiplexing method to realize communication. The traditional optimization method for realizing link weighting by adopting the cooperation of broadcasting routing requests for multiple times and communication handshake and a fixed algorithm has the obvious network collision and delay caused by information concurrence. Usually, the networking necessary information usually occupies most of the network performance, and the communication resources released to the users are extremely limited and have very low utilization rate. Meanwhile, due to lack of real-time monitoring of the states of devices and channels in the group network, communication path failure and target object product packet loss which cannot be found due to various environmental influences often occur, and network collision (such as zigbee) is caused.
Disclosure of Invention
Aiming at the technical problem, the invention provides a network optimization method for realizing dynamic weighting of communication links by using artificial intelligence, the optimization method obtains all link characteristics of communication paths of communication elements through the communication elements to record in real time and form a link table, and an intelligent visual network topological graph is generated according to the link characteristics of the link table; when a communication request exists, identifying an intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics among communication elements, introducing the link characteristics into a link image identification sample library for comparison, calculating a real-time dynamic weighted value of each link, calculating an optimal communication path according to the weighted value, and realizing communication according to the optimal communication path; and updating the intelligent visual network topological graph of the latest link characteristics participating in the communication and introducing the intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library so as to complete the optimization of the dynamic weighting of the communication network link.
The technical scheme of the invention is as follows:
a network optimization method for realizing dynamic weighting of communication links by artificial intelligence is characterized in that: the optimization method comprises the following steps:
s1: the communication elements participating in communication in the communication network acquire all link characteristics of the communication path of the communication elements to record in real time and form a link table, and an intelligent visual network topological graph is generated according to the link characteristics of the link table;
s2: when the communication element of the communication network has a communication request, the communication element initiating the communication request is an initiator, the communication element passing through the path is a forwarder, and the target communication element is an object; the initiator to the forwarder to the object is a communication path, and a link is formed between every two communication elements in the communication path; identifying the intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics between every two communication elements in all communication paths between an initiator and an object, importing the link characteristics into a link image identification sample library for comparison, and calculating a real-time dynamic weighted value of each link;
s3: calculating an optimal communication path according to the real-time dynamic weighted value of each link, and realizing communication according to the optimal communication path;
s4: after the communication is finished, updating the intelligent visual network topological graph by the latest link characteristics between every two communication elements in all communication paths between the initiator participating in the communication and the object; introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library, providing data support for link dynamic weighting and finishing optimization of communication network link dynamic weighting;
s5: waiting for the next communication request; when the communication element of the communication network has the latest communication request, the process returns to the step S1.
According to the method, all link characteristics of a communication path of a communication element are acquired through the communication element and recorded in real time to form a link table, and an intelligent visual network topological graph is generated according to the link characteristics of the link table; when a communication request exists, identifying an intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics among communication elements, introducing the link characteristics into a link image identification sample library for comparison, calculating a real-time dynamic weighted value of each link, calculating an optimal communication path according to the weighted value, and realizing communication according to the optimal communication path; and updating the intelligent visual network topological graph of the latest link characteristics participating in the communication and introducing the intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library so as to complete the optimization of the dynamic weighting of the communication network link.
Preferably: the link characteristics include, but are not limited to, initiator, object, location, link, channel, communication time slot, signal strength, communication elapsed time, historical success rate, type, and special capability.
Preferably: the intelligent visual network topological diagram is an intelligent visual network topological vector diagram which is generated by analyzing the link characteristics of the real-time communication states of all communication elements in the communication network at the same moment through the computer vision and image recognition technology and represents all the actual communication conditions through the combination of colors, sizes, shapes, line types, thicknesses and graphs, and is used for providing a uniform import and analysis data format for the computer vision and image recognition artificial intelligence technology.
Further: the updated intelligent visualization network topology map of step S4 is: the intelligent visual network topological graph automatically updates the vector diagrams which represent the communication paths and the link characteristics of all communication elements participating in communication through color, size, shape, line type, thickness and image combination according to the real-time dynamic weighted value of the link;
the invention identifies intelligent visual network topological diagram by computer vision and image identification technology, and obtains the link characteristic real-time situation of all communication elements participating in communication according to the color, size, shape, line type, thickness and graph combination of the topological diagram, and users can vividly master the concrete information of all communication elements by the topological diagram, such as: initiator, forwarder, object, communication path, map location of communication element, link between two communication elements, channel occupancy, communication time slot, signal strength, communication time consumption, historical success rate, type (base station, router, switch, cloud server), and special capability (computing capability, storage capability). After the communication is completed, the link with shortest communication time delay, fastest speed and minimum network overhead can be found out for communication by using the minimum nodes, the shortest path and the fastest path under the condition of minimum routing overhead, and after the communication is completed, the latest link characteristics participating in the communication are updated into an intelligent visual network topological graph; and introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library, providing data support for link dynamic weighting and finishing optimization of communication network link dynamic weighting.
And further: the link weighted value calculating method comprises the following steps:
the initiator of the communication element is source, the forwarder is jump, the object is target, the communication path of the communication element is link, and the link formed between every two communication elements is path; the weighted value is weight;
firstly, identifying all communication elements participating in communication marked by a solid line block diagram of the communication according to an intelligent visual network topological graph by using a computer vision and image identification technology;
then, identifying the link marked by the dotted line block diagram by using the same method, obtaining the link and identifying the link characteristics;
finally, the link characteristics of the communication are imported into a link image recognition sample library of the nerve convolution machine for comparison, and the weighted value weight of the links (path: source-jump 1), (path: jump1-jump …), (path: jump … -jump N) and (path: jump N-target) of the path link [ source, jump1, jump …, jump N and target ] is calculated.
Further, the method comprises the following steps: and the process of guiding the link characteristics participating in the communication into a link image identification sample library of the nerve convolution machine to perform contrast calculation of the link dynamic weight is as follows:
after the system introduces the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined, a comparison algorithm aiming at each link characteristic is represented by F (x), and the neural convolution calculation and comparison are carried out on the following link characteristics: f (communication element graphic size comparison), F (communication element graphic position comparison), F (communication element graphic combination comparison), F (communication element graphic color area comparison), F (link line shape comparison), F (link color comparison), F (link linear thickness comparison) and F (link-located area inequality risk assessment), namely, the link weighted value weight (path: source-jump) of the path link [ source, jump1, jump …, jump N and target ] can be calculated;
similarly, a link weight (path: source-jump 1) can be obtained;
similarly, the link weight (path: jump1-jump …);
similarly, the link weight (path: jump … -jump N) can be obtained;
similarly, the weight (path: jump N-target) of the link can be obtained;
after finishing the calculation and comparison of the path link [ source, jump1, jump …, jump N and target ], calculating other communication paths in the same way respectively to obtain the dynamic weight of each link under each communication path.
According to the invention, according to the calculation result of the weighted value of the link, the communication paths with more forwarders can be eliminated by the algorithm of the shortest node, and the optimal communication path can be obtained according to the accounting method of the dynamic link weighting; after each communication, a new visual topological graph is automatically generated according to the latest link characteristics, and after the communication is finished, the operation of learning the neural convolution machine is executed, so that continuous characteristic data can be generated for a communication network, a latest neural convolution machine learning sample library is formed, and data support is provided for dynamic weighting of links; the link image recognition sample library of the nerve convolution machine only focuses on link weighted values and link characteristics, and has good portability for networks with the same or similar network topologies; when the system leads the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined.
Preferably: before the step of S1, the following steps may be further included: the communication network is mainly formed by networking two or more communication elements, after the communication elements are routed and responded through broadcasting, the communication elements acquire corresponding communication channels and send and receive information in time slots to realize communication, and all the communication elements in the communication network complete networking.
Preferably: the communication elements include, but are not limited to: the system comprises a gateway and a node, wherein the gateway has computing and storage capacity.
Preferably: the gateway is a base station, a router, a switch or a cloud server.
In the network structure, if the communication between the nodes is generated, the initiator node firstly executes the sending according to the optimal path to the base station and then superposes the optimal path from the base station to the object node to complete the communication of the whole communication path because of the calculation and storage capacity of the base station. At this time, the advantages of the link dynamic weighting of the invention under the network delay are more highlighted; in the network structure, a plurality of base stations exist, the calculation storage capacity of the base stations can be distributed evenly or according to needs, the calculation storage pressure of a single base station is reduced, and the method is suitable for a distributed system; in the network structure, if the base station does not exist, the part of calculation and storage work can be transplanted to the cloud server, and the method is also suitable for the flexibly-arranged network topology structure without the base station.
By adopting the technical scheme, the invention has the beneficial effects that:
1) acquiring all link characteristics of a communication path of a communication element through the communication element, recording in real time, forming a link table, and generating an intelligent visual network topological graph according to the link characteristics of the link table; when a communication request exists, identifying an intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics among communication elements, introducing the link characteristics into a link image identification sample library for comparison, calculating a real-time dynamic weighted value of each link, calculating an optimal communication path according to the weighted value, and realizing communication according to the optimal communication path; and updating the intelligent visual network topological graph of the latest link characteristics participating in the communication and introducing the intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library so as to complete the optimization of the dynamic weighting of the communication network link.
2) The invention identifies intelligent visual network topological diagram by computer vision and image identification technology, and obtains the link characteristic real-time situation of all communication elements participating in communication according to the color, size, shape, line type, thickness and graph combination of the topological diagram, and users can vividly master the concrete information of all communication elements by the topological diagram, such as: initiator, forwarder, object, communication path, map location of communication element, link between two communication elements, channel occupancy, communication time slot, signal strength, communication time consumption, historical success rate, type (base station, router, switch, cloud server), and special capability (computing capability, storage capability). After the communication is completed, the link with shortest communication time delay, fastest speed and minimum network overhead can be found out for communication by using the minimum nodes, the shortest path and the fastest path under the condition of minimum routing overhead, and after the communication is completed, the latest link characteristics participating in the communication are updated into an intelligent visual network topological graph; and introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library, providing data support for link dynamic weighting and finishing optimization of communication network link dynamic weighting.
3) According to the invention, according to the calculation result of the weighted value of the link, the communication paths with more forwarders can be eliminated by the algorithm of the shortest node, and the optimal communication path can be obtained according to the accounting method of the dynamic link weighting; after each communication, a new visual topological graph is automatically generated according to the latest link characteristics, and after the communication is finished, the operation of learning the neural convolution machine is executed, so that continuous characteristic data can be generated for a communication network, a latest neural convolution machine learning sample library is formed, and data support is provided for dynamic weighting of links; the link image recognition sample library of the nerve convolution machine only focuses on link weighted values and link characteristics, and has good portability for networks with the same or similar network topologies; when the system leads the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined.
4.) the invention, in the network structure, if produce node and nodal communication, because the base transceiver station possesses calculation and storage ability, so the initiator node will carry out sending according to the optimum route to the base transceiver station first, then superpose the base transceiver station to the optimum route of the object node and finish the communication of the whole communication route. At this time, the advantages of the link dynamic weighting of the invention under the network delay are more highlighted; in the network structure, a plurality of base stations exist, the calculation storage capacity of the base stations can be distributed evenly or according to needs, the calculation storage pressure of a single base station is reduced, and the method is suitable for a distributed system; in the network structure, if the base station does not exist, the part of calculation and storage work can be transplanted to the cloud server, and the method is also suitable for the flexibly-arranged network topology structure without the base station.
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Drawings
FIG. 1 is a flow chart of a method for optimizing a network according to an embodiment;
FIG. 2 is a link table according to an embodiment;
FIG. 3 is a diagram illustrating an intelligent visual network topology of communication elements participating in communication in a communication network according to an embodiment;
FIG. 4 is a diagram of an intelligent visual network topology after networking is completed according to the embodiment;
FIG. 5 is an intelligent visual network topology graph generated according to link characteristics according to an embodiment;
FIG. 6 is a diagram of an intelligent visual network topology identified by computer vision and image recognition techniques according to an embodiment;
FIG. 7 is a schematic diagram of a neural network computing and comparing link dynamic weighting values for the neuro-convolution machine according to an embodiment;
fig. 8 is a schematic diagram of an example of introducing an updated intelligent visual network topology map into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1-8, the present invention discloses a network optimization method for implementing dynamic weighting of communication links by using artificial intelligence, wherein the optimization method comprises the following steps:
s1: the communication elements participating in communication in the communication network acquire all link characteristics of the communication path of the communication elements to record in real time and form a link table, and an intelligent visual network topological graph is generated according to the link characteristics of the link table;
s2: when the communication element of the communication network has a communication request, the communication element initiating the communication request is an initiator, the communication element passing through the path is a forwarder, and the target communication element is an object; the initiator to the forwarder to the object is a communication path, and a link is formed between every two communication elements in the communication path; identifying the intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics between every two communication elements in all communication paths between an initiator and an object, importing the link characteristics into a link image identification sample library for comparison, and calculating a real-time dynamic weighted value of each link;
s3: calculating an optimal communication path according to the real-time dynamic weighted value of each link, and realizing communication according to the optimal communication path;
s4: after the communication is finished, updating the intelligent visual network topological graph by the latest link characteristics between every two communication elements in all communication paths between the initiator participating in the communication and the object; introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library, providing data support for link dynamic weighting and finishing optimization of communication network link dynamic weighting;
s5: waiting for the next communication request; when the communication element of the communication network has the latest communication request, the process returns to the step S1.
According to the method, all link characteristics of a communication path of a communication element are acquired through the communication element and recorded in real time to form a link table, and an intelligent visual network topological graph is generated according to the link characteristics of the link table; when a communication request exists, identifying an intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics among communication elements, introducing the link characteristics into a link image identification sample library for comparison, calculating a real-time dynamic weighted value of each link, calculating an optimal communication path according to the weighted value, and realizing communication according to the optimal communication path; and updating the intelligent visual network topological graph of the latest link characteristics participating in the communication and introducing the intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library so as to complete the optimization of the dynamic weighting of the communication network link.
As shown in fig. 2, the link characteristics preferably include, but are not limited to, initiator, object, location, link, channel, communication time slot, signal strength, communication time consumption, historical success rate, type, and special capability.
As shown in fig. 6, preferably, the intelligent visual network topology map is an intelligent visual network topology vector map which is generated by analyzing link characteristics of real-time communication states of all communication elements in the communication network at the same time through computer vision and image recognition technology and represents all communication practical situations through color, size, shape, line type, thickness and graphic combination, and is used for providing a uniform import and analysis data format for computer vision and image recognition artificial intelligence technology.
Further, the updated intelligent visualization network topology map of step S4 is: the intelligent visual network topological graph automatically updates the vector diagrams which represent the communication paths and the link characteristics of all communication elements participating in communication through color, size, shape, line type, thickness and image combination according to the real-time dynamic weighted value of the link;
the invention identifies intelligent visual network topological diagram by computer vision and image identification technology, and obtains the link characteristic real-time situation of all communication elements participating in communication according to the color, size, shape, line type, thickness and graph combination of the topological diagram, and users can vividly master the concrete information of all communication elements by the topological diagram, such as: initiator, forwarder, object, communication path, map location of communication element, link between two communication elements, channel occupancy, communication time slot, signal strength, communication time consumption, historical success rate, type (base station, router, switch, cloud server), and special capability (computing capability, storage capability). After the communication is completed, the link with shortest communication time delay, fastest speed and minimum network overhead can be found out for communication by using the minimum nodes, the shortest path and the fastest path under the condition of minimum routing overhead, and after the communication is completed, the latest link characteristics participating in the communication are updated into an intelligent visual network topological graph; and introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library, providing data support for link dynamic weighting and finishing optimization of communication network link dynamic weighting.
As shown in fig. 6 and 7, the link weight value calculation method includes:
the initiator of the communication element is source, the forwarder is jump, the object is target, the communication path of the communication element is link, and the link formed between every two communication elements is path; the weighted value is weight;
firstly, identifying all communication elements participating in communication marked by a solid line block diagram of the communication according to an intelligent visual network topological graph by using a computer vision and image identification technology;
then, identifying the link marked by the dotted line block diagram by using the same method, obtaining the link and identifying the link characteristics;
finally, the link characteristics of the communication are imported into a link image recognition sample library of the nerve convolution machine for comparison, and the weighted value weight of the links (path: source-jump 1), (path: jump1-jump …), (path: jump … -jump N) and (path: jump N-target) of the path link [ source, jump1, jump …, jump N and target ] is calculated.
As shown in fig. 6 and fig. 7, the process of introducing the link characteristics participating in the communication into the link image recognition sample library of the neuro-convolution machine to perform the comparison calculation of the link dynamic weight:
after the system introduces the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined, a comparison algorithm aiming at each link characteristic is represented by F (x), and the neural convolution calculation and comparison are carried out on the following link characteristics: f (communication element graphic size comparison), F (communication element graphic position comparison), F (communication element graphic combination comparison), F (communication element graphic color area comparison), F (link line shape comparison), F (link color comparison), F (link linear thickness comparison) and F (link-located area inequality risk assessment), namely, the link weighted value weight (path: source-jump) of the path link [ source, jump1, jump …, jump N and target ] can be calculated;
similarly, a link weight (path: source-jump 1) can be obtained;
similarly, the link weight (path: jump1-jump …);
similarly, the link weight (path: jump … -jump N) can be obtained;
similarly, the weight (path: jump N-target) of the link can be obtained;
after finishing the calculation and comparison of the path link [ source, jump1, jump …, jump N and target ], calculating other communication paths in the same way respectively to obtain the dynamic weight of each link under each communication path.
According to the invention, according to the calculation result of the weighted value of the link, the communication paths with more forwarders can be eliminated by the algorithm of the shortest node, and the optimal communication path can be obtained according to the accounting method of the dynamic link weighting; after each communication, a new visual topological graph is automatically generated according to the latest link characteristics, and after the communication is finished, the operation of learning the neural convolution machine is executed, so that continuous characteristic data can be generated for a communication network, a latest neural convolution machine learning sample library is formed, and data support is provided for dynamic weighting of links; the link image recognition sample library of the nerve convolution machine only focuses on link weighted values and link characteristics, and has good portability for networks with the same or similar network topologies; when the system leads the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined.
As shown in fig. 1, 3 and 4, before the step of S1, the method may further include the following steps: the communication network is mainly formed by networking two or more communication elements, after the communication elements are routed and responded through broadcasting, the communication elements acquire corresponding communication channels and send and receive information in time slots to realize communication, and all the communication elements in the communication network complete networking.
As shown in fig. 6, the communication elements include, but are not limited to: the system comprises a gateway and a node, wherein the gateway has computing and storage capacity.
As shown in fig. 6, the gateway is a base station, a router, a switch, or a cloud server.
In the network structure, if the communication between the nodes is generated, the initiator node firstly executes the sending according to the optimal path to the base station and then superposes the optimal path from the base station to the object node to complete the communication of the whole communication path because of the calculation and storage capacity of the base station. At this time, the advantages of the link dynamic weighting of the invention under the network delay are more highlighted; in the network structure, a plurality of base stations exist, the calculation storage capacity of the base stations can be distributed evenly or according to needs, the calculation storage pressure of a single base station is reduced, and the method is suitable for a distributed system; in the network structure, if the base station does not exist, the part of calculation and storage work can be transplanted to the cloud server, and the method is also suitable for the flexibly-arranged network topology structure without the base station.
The first embodiment: as shown in fig. 3, a base station a and nodes b, c, d, e, f, g, h, i carrying one or more of the same communication technologies in a communication network are temporarily distributed in a certain area to form an irregular network topology, and any two of the base station a and the nodes b, c, d, e, f, g, h, i can be interconnected within an effective communication distance.
As shown in fig. 4, after the base station a broadcasts the routing information for the first time, the nodes b, c, d, e, f, g, h, i acquire the corresponding communication channels and time slots from the base station a through the response, and at this time, the networking process is completed.
As shown in fig. 2, the node can send and receive information in the allocated communication channel and timeslot to implement communication, and in each communication process, the base station a records the link characteristics of the nodes b, c, d, e, f, g, h, and i: the initiator, object, location, link, channel, communication time slot, signal strength, communication time consumption, historical success rate, type and special capability form a link table.
As shown in fig. 4, an intelligent visual network topological graph is generated through the link characteristic correspondence of the link table, wherein in the topological graph, the position distribution takes the position information of the base station a as the center, the polar coordinates of each other are calculated and displayed through the position information of the nodes b, c, d, e, f, g, h, i and the base station a, the map actual positions can be corresponded according to the scale conversion, and through the map actual position coordinates of the topological graph, the administrator can accurately grasp the specific positions of all communication elements, and the decision information help is provided for overhauling, moving and replacing the communication elements.
As shown in fig. 5, the base station a, the nodes b, c, d, e, f, g, h, i are represented by circles, and the special capabilities are distinguished by the combination of the circle radius and the color block, for example, the circle radius of the base station a is larger than that of other communication units, and represents that the base station a carries multiple communication technologies and multiple channels; the blue squares represent base station a with computing and storage capabilities. Each link represents communication time consumption in a thickness mode according to the distributed channels, represents on-off in a virtual-real mode according to signal strength and communication history success rate, and represents the occupation condition of the channels in a virtual-real mode according to time slots.
As shown in fig. 6, the base station a has computing and storage capabilities, multi-channel, multi-communication technology integration; the node has only single channel communication capability. Therefore, a large amount of calculation and analysis work is carried out by the base station a, and the node performs network communication according to the calculation result of the base station a.
Assuming that the base station a and the node d generate a communication request, the topology graph can analyze that the path thereof has four paths of link [ a, c, d ], [ a, h, d ], [ a, i, d ], [ a, f, c, d ].
As shown in fig. 6, the intelligent visual topological graph is introduced into the link image recognition sample library for comparison and calculation, and a dynamic link weighting value is obtained. The paths [ a, c, d ] are only taken as examples here, and the rest three paths are analogized.
It should be noted that the initiator of the communication element is source, the forwarder is jump, the object is target, the communication path of the communication element is link, and the link formed between every two communication elements is path; the weighted value is weight;
firstly, identifying all communication elements participating in communication marked by a solid line block diagram of the communication according to an intelligent visual network topological graph by using a computer vision and image identification technology;
then, the same method is used to identify the link marked by the dashed line block diagram, and the link path is obtained: a-c, path: c-d, and identifying link characteristics: initiator, object, location, link, channel, communication time slot, signal strength, communication time consumption, historical success rate, type, and special capability.
Finally, the link characteristics of the communication are imported into a link image recognition sample library of a nerve convolution machine for comparison, and the link path of a path link [ source, jump1, jump …, jump N and target ] is calculated: a-c, path: the weight of c-d is, for example, weight (path: a-c) =2 and weight (path: c-d) = 8.
As shown in fig. 7 and 8, to calculate the path: and a-c are taken as examples, and the process of guiding the link characteristics participating in the communication into a link image identification sample library of the nerve convolution machine to perform contrast calculation on the link dynamic weight.
After the system introduces the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined, a comparison algorithm aiming at each link characteristic is represented by F (x), and the neural convolution calculation and comparison are carried out on the following link characteristics: f (communication element figure size contrast), F (communication element figure position contrast), F (communication element figure combination contrast), F (communication element figure colour regional contrast), F (link linear contrast), F (link colour contrast), F (link linear thickness contrast), F (link regional ineligibility risk assessment), can calculate link: the weight of weight (path: a-c) of [ a, c, d ] path can be obtained similarly.
After the comparison of the paths [ a, c and d ] is finished, the dynamic weighted value of each link under each path of the results obtained by calculating other three paths in the same way is as follows:
Link：[a、c、d]、[weight（path：a-c）=2、weight（path：c-d）=8]；
Link：[a、h、d]、[weight（path：a-h）=6、weight（path：h-d）=7]；
Link：[a、i、d]、[weight（path：a-i）=3、weight（path：c-d）=8]；
Link：[a、f、c、d]、[weight（path：a-f）=0.3、weight（path：f-c）=1、weight（path：c-d）=8]。
according to the calculation result of the path, with the shortest node algorithm, Link: [ a, f, c and d ] are removed, and Link: [ a, h, d ] is the optimal path.
At this time, according to Link: the method comprises the following steps that [ a, h and d ] are optimal paths, communication between a base station a and a node d is executed, after the communication is finished, the node d records link characteristics of a path (a-h) and a path (h-d) in the communication, returns the link characteristics to the base station a through an original path, and updates an intelligent visual network topological graph; and the base station a introduces the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library.
It is worth noting that, each time communication occurs, a new visual network topological graph is automatically generated according to the latest data, and after the communication is finished, the operation of machine learning is executed, so that continuous characteristic data can be generated for the communication network, a latest link image recognition sample library is formed, and data support is provided for link dynamic weighting.
The base station and the nodes b, c, d, e, f, g, h and i form a network, and a communication process between the base station a and the node d is triggered once.
Second embodiment: in the network structure, if the node e communicates with the node d, the node e will perform transmission according to the optimal path to the node a and then superimpose the optimal path from the node a to the node d due to the calculation and storage capacity of the node a. At this time, the advantages of the link dynamic weighting of the present invention under the network delay are more prominent.
The third embodiment: if a plurality of base stations a exist in the network structure, the computing storage capacity of the base stations can be distributed evenly or according to needs, the computing storage pressure of a single base station is reduced, and the method is suitable for a distributed system.
The fourth embodiment: if the base station a does not exist, the part of calculation and storage work can be transplanted to a cloud server, and the method is also suitable for a flexibly-arranged network topology structure without the base station.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (9)
1. A network optimization method for realizing dynamic weighting of communication links by artificial intelligence is characterized in that: the network optimization method comprises the following steps:
s1: the communication elements participating in communication in the communication network acquire all link characteristics of the communication path of the communication elements to record in real time and form a link table, and an intelligent visual network topological graph is generated according to the link characteristics of the link table;
s2: when the communication element of the communication network has a communication request, the communication element initiating the communication request is an initiator, the communication element passing through the path is a forwarder, and the target communication element is an object; the initiator to the forwarder to the object is a communication path, and a link is formed between every two communication elements in the communication path; identifying the intelligent visual network topological graph through computer vision and image identification technology, thereby extracting link characteristics between every two communication elements in all communication paths between an initiator and an object, importing the link characteristics into a link image identification sample library for comparison, and calculating a real-time dynamic weighted value of each link;
s3: calculating an optimal communication path according to the real-time dynamic weighted value of each link, and realizing communication according to the optimal communication path;
s4: after the communication is finished, updating the intelligent visual network topological graph by the latest link characteristics between every two communication elements in all communication paths between the initiator participating in the communication and the object; introducing the updated intelligent visual network topological graph into a neural convolution machine learning library for machine learning to form a latest link image recognition sample library, providing data support for link dynamic weighting and finishing optimization of communication network link dynamic weighting;
s5: waiting for the next communication request; when the communication element of the communication network has the latest communication request, the process returns to the step S1.
2. The method of claim 1 for network optimization using artificial intelligence to achieve dynamic weighting of communication links, comprising: the link characteristics include, but are not limited to, initiator, object, location, link, channel, communication time slot, signal strength, communication elapsed time, historical success rate, type, and special capability.
3. The method of claim 2, further comprising the step of performing artificial intelligence to dynamically weight the communication links: the intelligent visual network topological diagram is an intelligent visual network topological vector diagram which is generated by analyzing the link characteristics of the real-time communication states of all communication elements in the communication network at the same moment through the computer vision and image recognition technology and represents all the actual communication conditions through the combination of colors, sizes, shapes, line types, thicknesses and graphs, and is used for providing a uniform import and analysis data format for the computer vision and image recognition artificial intelligence technology.
4. The method of claim 3 for network optimization using artificial intelligence to achieve dynamic weighting of communication links, comprising: the updated intelligent visualization network topology map of step S4 is: the intelligent visual network topological graph automatically updates the vector diagrams which represent the communication paths and the link characteristics of all communication elements participating in communication through color, size, shape, line type, thickness and image combination according to the real-time dynamic weighted value of the link;
according to the updated vector diagram, the link with shortest communication time delay, fastest speed and minimum network overhead can be found out for communication by the least nodes, the shortest path and the fastest path under the condition of minimum route overhead.
5. The method of claim 4 for network optimization using artificial intelligence to achieve dynamic weighting of communication links, comprising: the link weighted value calculating method comprises the following steps:
the initiator of the communication element is source, the forwarder is jump, the object is target, the communication path of the communication element is link, and the link formed between every two communication elements is path; the weighted value is weight;
firstly, identifying all communication elements participating in communication marked by a solid line block diagram of the communication according to an intelligent visual network topological graph by using a computer vision and image identification technology;
then, identifying the link marked by the dotted line block diagram by using the same method, obtaining the link and identifying the link characteristics;
finally, the link characteristics of the communication are imported into a link image recognition sample library of the nerve convolution machine for comparison, and the weighted value weight of the links (path: source-jump 1), (path: jump1-jump …), (path: jump … -jump N) and (path: jump N-target) of the path link [ source, jump1, jump …, jump N and target ] is calculated.
6. The method of claim 5 for network optimization using artificial intelligence to achieve dynamic weighting of communication links, comprising: and the process of guiding the link characteristics participating in the communication into a link image identification sample library of the nerve convolution machine to perform contrast calculation of the link dynamic weight is as follows:
after the system introduces the link characteristics of the link into the neural network of the multilayer neural convolution machine, the complexity of the layer number vision network of the neural network and the number of characteristic points are determined, a comparison algorithm aiming at each link characteristic is represented by F (x), and the neural convolution calculation and comparison are carried out on the following link characteristics: f (communication element graph size comparison), F (communication element graph position comparison), F (communication element graph combination comparison), F (communication element graph color area comparison), F (link line shape comparison), F (link color comparison), F (link line thickness comparison) and F (link area inelasticity risk assessment);
then the link weighted value weight (path: source-jump) of the path link [ source, jump1, jump …, jump N, target ] can be calculated;
similarly, a link weight (path: source-jump 1) can be obtained;
similarly, the link weight (path: jump1-jump …);
similarly, the link weight (path: jump … -jump N) can be obtained;
similarly, the weight (path: jump N-target) of the link can be obtained;
after finishing the calculation and comparison of the path link [ source, jump1, jump …, jump N and target ], calculating other communication paths in the same way respectively to obtain the dynamic weight of each link under each communication path.
7. The method of any of claims 1-6 for network optimization using artificial intelligence to achieve dynamic weighting of communication links, comprising: before the step of S1, the following steps may be further included: the communication network is mainly formed by networking two or more communication elements, after the communication elements are routed and responded through broadcasting, the communication elements acquire corresponding communication channels and send and receive information in time slots to realize communication, and all the communication elements in the communication network complete networking.
8. The method of any of claims 1-6 for network optimization using artificial intelligence to achieve dynamic weighting of communication links, comprising: the communication elements include, but are not limited to: a gateway and a node.
9. The method of claim 8, further comprising the step of performing artificial intelligence to dynamically weight the communication links: the gateway is a base station, a router, a switch or a cloud server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010521746.2A CN111683384B (en) | 2020-06-10 | 2020-06-10 | Network optimization method for realizing dynamic weighting of communication link by artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010521746.2A CN111683384B (en) | 2020-06-10 | 2020-06-10 | Network optimization method for realizing dynamic weighting of communication link by artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111683384A true CN111683384A (en) | 2020-09-18 |
| CN111683384B CN111683384B (en) | 2023-01-24 |
Family
ID=72454484
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010521746.2A Active CN111683384B (en) | 2020-06-10 | 2020-06-10 | Network optimization method for realizing dynamic weighting of communication link by artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111683384B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112968798A (en) * | 2021-02-04 | 2021-06-15 | 西安热工研究院有限公司 | Virtual and real network environment automatic generation method suitable for electric power industrial control target range platform |
| CN114745763A (en) * | 2022-03-08 | 2022-07-12 | 深圳市欣博跃电子有限公司 | Method and system for updating optimal network topology |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130044755A1 (en) * | 2011-08-16 | 2013-02-21 | Futurewei Technologies, Inc. | Scalable Packet Scheduling Policy for Vast Number of Sessions |
| CN104010343A (en) * | 2014-06-11 | 2014-08-27 | 沈阳中科博微自动化技术有限公司 | Wireless network optimization method for intelligent meter reading system |
| CN106879071A (en) * | 2017-02-13 | 2017-06-20 | 电子科技大学 | Wireless Mesh netword low time delay optimization method based on dynamic time slot allocation |
| US20200154369A1 (en) * | 2006-10-03 | 2020-05-14 | Interdigital Technology Corporation | Combined open loop/closed loop (cqi-based) uplink transmit power control with interference mitigation for e-utra |
-
2020
- 2020-06-10 CN CN202010521746.2A patent/CN111683384B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200154369A1 (en) * | 2006-10-03 | 2020-05-14 | Interdigital Technology Corporation | Combined open loop/closed loop (cqi-based) uplink transmit power control with interference mitigation for e-utra |
| US20130044755A1 (en) * | 2011-08-16 | 2013-02-21 | Futurewei Technologies, Inc. | Scalable Packet Scheduling Policy for Vast Number of Sessions |
| CN104010343A (en) * | 2014-06-11 | 2014-08-27 | 沈阳中科博微自动化技术有限公司 | Wireless network optimization method for intelligent meter reading system |
| CN106879071A (en) * | 2017-02-13 | 2017-06-20 | 电子科技大学 | Wireless Mesh netword low time delay optimization method based on dynamic time slot allocation |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112968798A (en) * | 2021-02-04 | 2021-06-15 | 西安热工研究院有限公司 | Virtual and real network environment automatic generation method suitable for electric power industrial control target range platform |
| CN112968798B (en) * | 2021-02-04 | 2023-02-07 | 西安热工研究院有限公司 | Virtual and real network environment automatic generation method suitable for electric power industrial control target range platform |
| CN114745763A (en) * | 2022-03-08 | 2022-07-12 | 深圳市欣博跃电子有限公司 | Method and system for updating optimal network topology |
| CN114745763B (en) * | 2022-03-08 | 2023-08-22 | 深圳市欣博跃电子有限公司 | Method and system for updating optimal network topology |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111683384B (en) | 2023-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kuklinski et al. | Density based clustering algorithm for vehicular ad-hoc networks | |
| CN111683384B (en) | Network optimization method for realizing dynamic weighting of communication link by artificial intelligence | |
| WO2004110082B1 (en) | System and method for determining location of a device in a wireless communication network | |
| US20090285136A1 (en) | Virtual Grid | |
| CN108683445B (en) | Aircraft formation layered network access management method and system | |
| CN110650039A (en) | Multimodal optimization-based network collaborative communication model for unmanned aerial vehicle cluster-assisted vehicle | |
| CN111556550B (en) | Routing method for unmanned aerial vehicle network communication | |
| CN110213175B (en) | Intelligent management and control system and method for knowledge-defined network | |
| CN109275154B (en) | Dynamic self-adaptive routing path planning method based on double-layer topological routing algorithm | |
| CN110519020A (en) | Unmanned systems network intelligence cross-layer data transmission method and system | |
| CN108667668B (en) | Accessibility routing method based on Internet of vehicles communication base in urban road scene | |
| CN111405634B (en) | Method and device for self-adaptive clustering of wireless sensor network | |
| CN113411766B (en) | Intelligent Internet of things comprehensive sensing system and method | |
| CN107819686B (en) | Unmanned aerial vehicle ad hoc network geographic information routing method and system based on position prediction | |
| CN109587751A (en) | Unmanned plane ad hoc network adaptive routing method and system | |
| CN107222901B (en) | Method for realizing cognitive radio network routing protocol based on channel allocation | |
| CN109152079B (en) | Time slot scheduling and channel allocation method for TSCH (three-channel switched capacitor) network | |
| Meng et al. | Intelligent routing orchestration for ultra-low latency transport networks | |
| CN115665860A (en) | Unmanned aerial vehicle ad hoc network resource allocation method based on migratory bird swarm characteristics | |
| CN109474523B (en) | Networking method and system based on SDN | |
| CN115118647A (en) | System and method for perceiving and announcing computing power information in computing power network | |
| CN114641049A (en) | Unmanned aerial vehicle ad hoc network layered routing method based on fuzzy logic | |
| CN111615203B (en) | Task-driven joint channel time slot allocation method facing data center | |
| HaghighiFard et al. | Hierarchical Federated Learning in Multi-hop Cluster-Based VANETs | |
| CN110493068A (en) | A kind of network route generating method and system |
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 |