CN117336226A - Path optimization method based on combination of elastic communication network and AI algorithm - Google Patents
Path optimization method based on combination of elastic communication network and AI algorithm Download PDFInfo
- Publication number
- CN117336226A CN117336226A CN202311409864.4A CN202311409864A CN117336226A CN 117336226 A CN117336226 A CN 117336226A CN 202311409864 A CN202311409864 A CN 202311409864A CN 117336226 A CN117336226 A CN 117336226A
- Authority
- CN
- China
- Prior art keywords
- path
- weight
- link
- algorithm
- node
- 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.)
- Pending
Links
- 238000004891 communication Methods 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000005457 optimization Methods 0.000 title claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims abstract description 39
- 230000007246 mechanism Effects 0.000 claims description 41
- 230000006855 networking Effects 0.000 claims description 38
- 238000013507 mapping Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 6
- 238000007405 data analysis Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/02—Topology update or discovery
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/123—Evaluation of link metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/124—Shortest path evaluation using a combination of metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/125—Shortest path evaluation based on throughput or bandwidth
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/14—Routing performance; Theoretical aspects
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a path optimization method based on an elastic communication network and combining an AI algorithm, which is characterized in that after an elastic communication network operation system receives a service request and forwards data and marks, the performance of equipment and links, network statistical information and the like are used as references, weights of the links in the elastic communication network are calculated, a topological graph with weights is generated, and path calculation is carried out in the topological graph by searching network positions of a flow starting point and a flow end point to obtain a plurality of selectable routing paths and corresponding path weights; after a plurality of selectable routing paths and weights thereof are obtained, a load balancing strategy is applied to the selectable routing paths, and a part of the selectable routing paths is selected as a preferred routing path set. Meanwhile, in order to meet different path calculation requirements of the elastic communication network, the operation system of the elastic communication network adopts a plurality of path calculation methods to meet the path selection requirements while keeping a basic shortest path algorithm.
Description
Technical Field
The invention belongs to the field of elastic communication networks, and particularly relates to a path optimization method based on an elastic communication network combined with an AI algorithm.
Background
In the sky-ground three-dimensional elastic communication network, various network running state information from multiple channels and multiple dimensions of the sky-ground is collected in real time, fusion of multi-source and multi-dimensional information is carried out, an OODA dynamic cognitive ring can be triggered, and finally a decision is generated and executed. The network security event prediction mainly refers to the application of scientific theory, method and existing experience to judge and predict the development trend and hazard condition of the important security event found in the network system, and is an important stage of network security situation awareness, and the main goal of network security situation awareness is to predict the network security event. The current network system has a plurality of services, and the network functions are continuously expanded, so that the security factors influencing the network security situation are more and more, and the problem that the current elastic communication network field needs to be solved is that various complex association relations exist before various factors so that comprehensive perception information is difficult to obtain.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a wireless data compression optimization method based on cooperative spectrum sensing.
The technical scheme of the invention is as follows:
a method for combining AI algorithm path optimization based on an elastic communication network, comprising the following steps:
(1) Establishing a networking control mechanism; the networking control mechanism comprises a flow triggering networking control mechanism, a pre-routing networking control mechanism and a service demand triggering networking control mechanism; the data forwarding is carried out through three mechanisms, and the adopted networking control mechanism is marked, so that basic data is provided for subsequent data analysis;
(2) Analyzing the networking control mechanism marks carried in the data forwarded in the step (1), and calculating different forwarding paths according to different networking control mechanisms; after receiving the service request, forwarding the data and the mark, the elastic communication network operating system calculates the weight of the link in the elastic communication network by taking the performance of the equipment and the link and the network statistical information as references, and generates a topological graph with the weight;
(3) The method comprises the steps of performing path calculation in a topological graph by searching network positions of a flow starting point and a flow end point to obtain a plurality of selectable routing paths and corresponding path weights;
(4) After a plurality of selectable routing paths and weights thereof are obtained, a load balancing strategy is applied to the selectable routing paths, and a part of the selectable routing paths is selected as a preferred routing path set.
For further explanation of the present invention, the flow triggering networking control mechanism marks that the received data message is reported to the elastic communication network operating system through the southbound interface protocol to perform the processing procedures of analyzing the data forwarding requirement, path calculation and strategy issuing to realize data forwarding and marking; the pre-routing networking control mechanism marks that when data arrives, the elastic communication network node equipment carries out high-speed forwarding according to the pre-prepared flow table information; and the service demand triggering networking control mechanism marks that when data arrives, the equipment performs feature matching on the service data and performs forwarding meeting the service quality.
To further explain the present invention, the method further includes first, considering whether the path-finding node is associated with the core service node when traversing the nodes of the full network topology graph, and preferentially recording nodes intersected with the core node or links through which the core node data passes when each node is experienced, so as to generate a trust level mark; secondly, integrating all the node and link information recorded at the moment into a selectable route path, and recording a result set; then, if the node which has undergone before reaching is found in the period, if the trust level mark exists, a routing loop is judged to be formed, and backtracking is carried out upwards; finally, a plurality of loop-free paths from the source host to the destination host are obtained and are selectable routing paths based on the trust degree. The route selection is more accurate through the step, and the problem of abnormal route caused by insufficient trust is avoided.
For further explanation of the present invention, the calculating further includes calculating a bandwidth weight for each of the selectable routing paths; the method comprises the following specific steps: firstly, calculating the weight of each link on a path, acquiring the current working bandwidth of the link, acquiring the current residual bandwidth of the link, and calculating the weight through the following formula:
weight = 100- (remaining link bandwidth +.current link bandwidth x 100);
secondly, selecting the maximum value of the weights of all links on the path as the bandwidth weight of the path; because the loads of the links are different, the link with the greatest load will become a short board of the entire path, depending on the barrel effect.
For further explanation of the present invention, the method further includes dividing the bandwidth by the weight of each link on the path to use the calculation index as another weight reference index, and the specific steps are as follows:
first, a calculation force weight model is constructed:
wherein C is br Is the total calculation force requirement; f (x) is a mapping function; alpha, beta and gamma are mapping proportionality coefficients; q 1 、q 2 、q 3 Calculating force for redundancy; the summation operation from j to m is used for summing the values of the mapping functions representing the parallel computing capacities which can be provided by m parallel computing chips b, wherein a, b and c are used for distinguishing chips of different types, m, n and p are used for distinguishing the respective quantity summation of the chips of different types, i, j and k are counting mark variables for recording the chips of different types, and only significance exists in a summation formula;
taking the parallel computing capability as an example, assuming that b1, b2, b3 3 different types of parallel computing chip resources exist, f (b) j ) Mapping function, q, representing parallel computing power available to jth parallel computing chip b 2 Representing redundant computational power of parallel computation;
calculation force weight = 100- (idle link calculation force +.current link calculation force +.100)
And secondly, selecting the maximum value of the calculation force weights of all links on the path as the calculation force weight of the path. Because the computational load varies from link to link, the link with the greatest computational load will become a short board of the entire path, depending on the barrel effect.
And finally, comprehensively judging the optimal weight path according to the algorithm weight and the bandwidth weight sequence. Therefore, the situation that the intelligent situation awareness service is not supported due to insufficient node operation capability caused by the fact that the weight standard is carried out through the bandwidth is avoided.
For further explanation of the present invention, by comparing the bandwidth weights of the paths, a set of paths with smaller weights is selected as the preferred routing path; the weight of at least one path in the group of paths is the minimum value in the weights of all the selectable routing paths; and after the weights of the rest preferred routing paths are differed from the weights of the paths, the difference is within a preset tolerance range.
And obtaining the routing path with the load balancing significance through the preferred steps. At this point, other attribute constraints may be selected for further routing. This step is the optimization of the routing result, and is also where the multi-attribute constraints jointly decide.
For further explanation of the present invention, the links between the devices of the resilient communication network node are abstracted to Link, and the links from the devices to the end hosts are abstracted to EdgeLink; in the obtained topology diagram, the last kilometer edge link is not included, so that two edge links are added to the result of the whole preferred route path obtained by the topology diagram to form a complete path, and then a path decision is issued to the network.
For further explanation of the present invention, the selectable routing paths calculate the shortest path by combining a k-shortest path algorithm and a shortest disjoint path algorithm, surbanle algorithm.
For further explanation of the present invention, the k-shortest path algorithm adopts a deviation path algorithm, specifically: first find from source point s to topology GThe shortest path of the target point t is taken as a first path p i ＝{p i [0]，p i [1]… }; the method comprises the steps of carrying out a first treatment on the surface of the Then delete path p in turn i Each node p i [j]Delete edge (p) i [j]，p i [j+1]) Then find p in the remaining topology G i [j]The shortest path to t is then added to the candidate path set { C ij In }; wherein to avoid ring generation in the path, p i [j]The new path to t is queried on the remaining topology graph G'; finally in these candidate path sets { C ij Finding a shortest path from s to t as the next shortest path; the entire process loops in turn until the first k shortest paths are found or no new shortest paths are generated.
For further explanation of the present invention, the surbanle algorithm calculates the shortest disjoint path from the source node to the destination node, and may calculate the working path and the protection path simultaneously, which is specifically as follows:
(1) The network topology can be abstracted as G (V, L, F), where V is the combination of nodes in the network, L is the set of links, F is the set of certain available resources, the traffic demand is expressed as R (s, d, w), (s, d) is the source-destination node pair, and w is the transmission demand for a certain resource; in graph G (V, L, F), a Dijkstra algorithm is used to calculate a shortest path tree T from source node s, the tree T containing the shortest paths from s to any node u; setting P as the shortest path from s to t;
(2) Updating each link weight in graph G with the notation w (u, v) =w (u, v) -d (u, v) +d (s, u); wherein w (u, v) represents the weight of a link (u, v) between node u and node v, the weight being related to a certain resource size of the link, the more resources the smaller the weight; d (s, v) + is the path weight of node s to node v; after updating, the link weight in the tree T is 0, and the link weight in the non-tree T is non-negative;
(3) Removing the link on the P path in G, and updating the reverse path link weight corresponding to P to 0;
(4) The shortest path P2 from the source node s to the destination node t in the graph G is updated after being calculated by using the Dijkstra algorithm again;
(5) And deleting the shared link segment of P and P2 in the graph G, and combining the rest links into two paths, namely the two required separated paths.
The invention has the beneficial effects that:
according to the method, when data arrives, the data is forwarded through three mechanisms, and the adopted networking control mechanism is marked, so that basic data is provided for subsequent data analysis; secondly, comprehensively calculating a shortest path by adopting a k shortest path algorithm and a shortest disjoint path algorithm-Suurballer algorithm; in order to meet different path calculation requirements of the elastic communication network, the operation system of the elastic communication network adopts a plurality of path calculation methods to meet the path selection requirements while keeping a basic shortest path algorithm, so that the stability and the accuracy of the method are further improved;
the method has the specific advantages that:
(1) The shortest path algorithm can calculate the shortest path reaching the target node, so that the time and cost of data packet transmission in the network are minimized, the transmission efficiency and response speed of the network can be improved by selecting the shortest path, and the Internet surfing experience is optimized;
(2) The shortest path algorithm can dynamically select an optimal path according to the real-time state and the load condition of the network, so that the load balance is realized, and the throughput and the performance stability of the network can be improved by reasonably distributing network resources and traffic;
(3) The shortest path algorithm can automatically detect faults in the network, quickly calculate alternative paths, realize fault recovery and reliability of the network, and can quickly adjust routes when network faults occur, so that normal transmission of data and availability of services are ensured;
(4) The method can process the graph with the negative weight edge by adopting an algorithm, has smaller coding complexity and good expansibility, can adapt to a plurality of problems after expansion, can automatically detect faults in a network, can rapidly calculate an alternative path, and realizes fault recovery and network reliability.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
Detailed Description
The invention will be further illustrated with reference to specific examples.
Examples:
a method for combining AI algorithm path optimization based on an elastic communication network, which is characterized in that: the method comprises the following steps:
(1) Establishing a networking control mechanism; the networking control mechanism comprises a flow triggering networking control mechanism, a pre-routing networking control mechanism and a service demand triggering networking control mechanism; the data forwarding is carried out through three mechanisms, and the adopted networking control mechanism is marked, so that basic data is provided for subsequent data analysis;
(2) Analyzing the networking control mechanism marks carried in the data forwarded in the step (1), and calculating different forwarding paths according to different networking control mechanisms; after receiving the service request, forwarding the data and the mark, the elastic communication network operating system calculates the weight of the link in the elastic communication network by taking the performance of the equipment and the link and the network statistical information as references, and generates a topological graph with the weight;
(3) The method comprises the steps of performing path calculation in a topological graph by searching network positions of a flow starting point and a flow end point to obtain a plurality of selectable routing paths and corresponding path weights;
(4) After a plurality of selectable routing paths and weights thereof are obtained, a load balancing strategy is applied to the selectable routing paths, and a part of the selectable routing paths is selected as a preferred routing path set.
The flow triggering networking control mechanism marks that the received data message is reported to an elastic communication network operation system through a southbound interface protocol to realize data forwarding and marking by analyzing the data forwarding requirement, calculating a path and issuing a strategy;
the flow triggering mechanism refers to how a certain data flow in the network is forwarded, the elastic communication network node equipment is unknown at first, and the elastic communication network node equipment needs to receive a data message and report the data message to an elastic communication network operation system through a southbound interface protocol to perform processing procedures such as analysis, path calculation, strategy issuing and the like of the data forwarding requirement so as to realize forwarding of the data flow;
the method comprises the steps that a flow triggers a networking control process, a certain data message arrives at an accessed elastic communication network node device, forwarding rules for the data message are not provided on an initial elastic communication network node device, the message is reported to an elastic communication network operation system, the elastic communication network operation system calculates a forwarding path through network topology resource information collected before, a forwarding flow table of a device along the way is generated according to a calculated forwarding path result, the forwarding flow table is issued to corresponding devices through a southbound interface protocol, and the devices control data according to the flow table to realize high-speed forwarding.
The pre-routing networking control mechanism marks that when data arrives, the elastic communication network node equipment carries out high-speed forwarding according to the pre-prepared flow table information;
the pre-routing networking control mechanism refers to that after network planning is started, an elastic communication network operating system issues a pre-planned flow table in advance according to network planning address information to realize whole network intercommunication, and then re-planning is carried out on pre-planned flow table information only according to state changes such as access termination, network nodes or links and the like;
the elastic communication network operation system in the pre-routing networking control process firstly receives address planning information of the network, collects information such as network source information, packet network state, access terminals and the like, performs forwarding path calculation of full network intercommunication, and sends a forwarding flow table to each elastic communication network node device in the network in advance according to a calculation result, and when data arrives, the elastic communication network node device performs high-speed forwarding according to the information of the pre-made flow table.
The service demand triggering networking control mechanism marks that when data arrives, the equipment performs feature matching on service data and performs forwarding meeting the service quality;
the service demand triggering networking control mechanism is that an elastic communication network operating system receives service forwarding demands, performs path calculation according to the service demands and the current state of the network, and issues a forwarding flow table to realize the on-demand forwarding of the service. The process differs from the stream triggering in that the resilient communication network operating system only interacts with the traffic control messages, but not with the traffic data;
the method comprises the steps that in the process of triggering service requirements, an elastic communication network operation system firstly receives service requirement information through a northbound interface, the information comprises service feature description, service quality guarantee requirements of service on bandwidth, time delay, jitter and the like, forwarding purposes of the service and the like, after the elastic communication network operation system receives the service requirement information, constraint path calculation meeting the service requirements is carried out according to the current network state, forwarding flow tables and service quality guarantee requirements are issued to a path elastic communication network node device according to calculation results, the device carries out QoS (quality of service) operation and the like according to the information such as the flow tables and the service quality guarantee requirements, the device needs to distribute to channel devices, and when data arrives, the device carries out feature matching on the service data and forwards the service data meeting the service quality.
The method further comprises the steps that firstly, whether a path-finding node is related to a core service node is considered when traversing nodes of the full-network topological graph, and when each node is experienced, nodes intersected with the core node or links through which the core node data passes are recorded preferentially, so that a trust degree mark is generated; secondly, integrating all the node and link information recorded at the moment into a selectable route path, and recording a result set; then, if the node which has undergone before reaching is found in the period, if the trust level mark exists, a routing loop is judged to be formed, and backtracking is carried out upwards; finally, a plurality of loop-free paths from the source host to the destination host are obtained and are selectable routing paths based on the trust degree. The route selection is more accurate through the step, and the problem of abnormal route caused by insufficient trust is avoided.
Calculating bandwidth weight of each selectable route path; the method comprises the following specific steps: firstly, calculating the weight of each link on a path, acquiring the current working bandwidth of the link, acquiring the current residual bandwidth of the link, and calculating the weight through the following formula:
weight = 100- (remaining link bandwidth +.current link bandwidth x 100);
secondly, selecting the maximum value of the weights of all links on the path as the bandwidth weight of the path; because the loads of the links are different, the link with the greatest load will become a short board of the entire path, depending on the barrel effect.
The method also comprises dividing the bandwidth of the weight of each link on the path by taking the calculation index as another weight reference index, and comprises the following specific steps:
first, a calculation force weight model is constructed:
wherein C is br Is the total calculation force requirement; f (x) is a mapping function; alpha, beta and gamma are mapping proportionality coefficients; q 1 、q 2 、q 3 Calculating force for redundancy; the summation operation from j to m is used for summing the values of the mapping functions representing the parallel computing capacities which can be provided by m parallel computing chips b, wherein a, b and c are used for distinguishing chips of different types, m, n and p are used for distinguishing the respective quantity summation of the chips of different types, i, j and k are counting mark variables for recording the chips of different types, and only significance exists in a summation formula;
taking the parallel computing capability as an example, assuming that b1, b2, b3 3 different types of parallel computing chip resources exist, f (b) j ) Mapping function, q, representing parallel computing power available to jth parallel computing chip b 2 Representing redundant computational power of parallel computation;
calculation force weight = 100- (idle link calculation force +.current link calculation force +.100)
And secondly, selecting the maximum value of the calculation force weights of all links on the path as the calculation force weight of the path. Because the computational load varies from link to link, the link with the greatest computational load will become a short board of the entire path, depending on the barrel effect.
And finally, comprehensively judging the optimal weight path according to the algorithm weight and the bandwidth weight sequence. Therefore, the situation that the intelligent situation awareness service is not supported due to insufficient node operation capability caused by the fact that the weight standard is carried out through the bandwidth is avoided.
The bandwidth weight of each path is compared, and a group of paths with smaller weight are selected as preferred routing paths; the weight of at least one path in the group of paths is the minimum value in the weights of all the selectable routing paths; and after the weights of the rest preferred routing paths are differed from the weights of the paths, the difference is within a preset tolerance range.
And obtaining the routing path with the load balancing significance through the preferred steps. At this point, other attribute constraints may be selected for further routing. This step is the optimization of the routing result, and is also where the multi-attribute constraints jointly decide.
The Link between the elastic communication network node devices is abstracted to be Link, and the Link from the device to the terminal host is abstracted to be EdgeLink; in the obtained topology diagram, the last kilometer edge link is not included, so that two edge links are added to the result of the whole preferred route path obtained by the topology diagram to form a complete path, and then a path decision is issued to the network.
The selectable routing path comprehensively calculates the shortest path by adopting a k shortest path algorithm and a shortest disjoint path algorithm Suurballer algorithm.
The k-shortest path algorithm adopts a deviation path algorithm, and specifically comprises the following steps: first, the shortest path from the source point s to the target point t is found in the topology map G as the first path p i ＝{p i [0]，p i [1]… }; the method comprises the steps of carrying out a first treatment on the surface of the Then delete path p in turn i Each node p i [j]Delete edge (p) i [j]，p i [j+1]) Then find p in the remaining topology G i [j]The shortest path to t is then added to the candidate path set { C ij In }; wherein to avoid ring generation in the path, p i [j]The new path to t is queried on the remaining topology graph G'; finally in these candidate path sets { C ij Finding a shortest path from s to t as the next shortest path; the whole process loops in turn until the first k shortest paths or paths are foundNo new shortest path occurs.
The Suurbelle algorithm calculates the shortest disjoint path from the source node to the destination node, and can calculate the working path and the protection path at the same time, and the method specifically comprises the following steps:
(1) The network topology can be abstracted as G (V, L, F), where V is the combination of nodes in the network, L is the set of links, F is the set of certain available resources, the traffic demand is expressed as R (s, d, w), (s, d) is the source-destination node pair, and w is the transmission demand for a certain resource; in graph G (V, L, F), a Dijkstra algorithm is used to calculate a shortest path tree T from source node s, the tree T containing the shortest paths from s to any node u; setting P as the shortest path from s to t;
(2) Updating each link weight in graph G with the notation w (u, v) =w (u, v) -d (u, v) +d (s, u); wherein w (u, v) represents the weight of a link (u, v) between node u and node v, the weight being related to a certain resource size of the link, the more resources the smaller the weight; d (s, v) + is the path weight of node s to node v; after updating, the link weight in the tree T is 0, and the link weight in the non-tree T is non-negative;
(3) Removing the link on the P path in G, and updating the reverse path link weight corresponding to P to 0;
(4) The shortest path P2 from the source node s to the destination node t in the graph G is updated after being calculated by using the Dijkstra algorithm again;
(5) And deleting the shared link segment of P and P2 in the graph G, and combining the rest links into two paths, namely the two required separated paths.
Claims (10)
1. A method for combining AI algorithm path optimization based on an elastic communication network, which is characterized in that: the method comprises the following steps:
(1) Establishing a networking control mechanism; the networking control mechanism comprises a flow triggering networking control mechanism, a pre-routing networking control mechanism and a service demand triggering networking control mechanism; the data forwarding is carried out through three mechanisms, and the adopted networking control mechanism is marked, so that basic data is provided for subsequent data analysis;
(2) Analyzing the networking control mechanism marks carried in the data forwarded in the step (1), and calculating different forwarding paths according to different networking control mechanisms; after receiving the service request, forwarding the data and the mark, the elastic communication network operating system calculates the weight of the link in the elastic communication network by taking the performance of the equipment and the link and the network statistical information as references, and generates a topological graph with the weight;
(3) The method comprises the steps of performing path calculation in a topological graph by searching network positions of a flow starting point and a flow end point to obtain a plurality of selectable routing paths and corresponding path weights;
(4) After a plurality of selectable routing paths and weights thereof are obtained, a load balancing strategy is applied to the selectable routing paths, and a part of the selectable routing paths is selected as a preferred routing path set.
2. The method for combining AI algorithm path optimization based on elastic communication network according to claim 1, wherein: the flow triggering networking control mechanism marks that the received data message is reported to an elastic communication network operation system through a southbound interface protocol to realize data forwarding and marking by analyzing the data forwarding requirement, calculating a path and issuing a strategy; the pre-routing networking control mechanism marks that when data arrives, the elastic communication network node equipment carries out high-speed forwarding according to the pre-prepared flow table information; and the service demand triggering networking control mechanism marks that when data arrives, the equipment performs feature matching on the service data and performs forwarding meeting the service quality.
3. The method for combining AI algorithm path optimization based on elastic communication network according to claim 1, wherein: the method further comprises the steps that firstly, whether a path-finding node is related to a core service node is considered when traversing nodes of the full-network topological graph, and when each node is experienced, nodes intersected with the core node or links through which the core node data passes are recorded preferentially, so that a trust degree mark is generated; secondly, integrating all the node and link information recorded at the moment into a selectable route path, and recording a result set;
then, if the node which has undergone before reaching is found in the period, if the trust level mark exists, a routing loop is judged to be formed, and backtracking is carried out upwards; finally, a plurality of loop-free paths from the source host to the destination host are obtained and are selectable routing paths based on the trust degree.
4. A method for combining AI algorithm path optimization based on elastic communication network as set forth in claim 3, wherein: calculating bandwidth weight of each selectable route path; the method comprises the following specific steps: firstly, calculating the weight of each link on a path, acquiring the current working bandwidth of the link, acquiring the current residual bandwidth of the link, and calculating the weight through the following formula:
weight = 100- (remaining link bandwidth +.current link bandwidth x 100);
and secondly, selecting the maximum value of the weights of all links on the path as the bandwidth weight of the path.
5. The method for combining AI algorithm path optimization based on elastic communication network as set forth in claim 4, wherein: the method also comprises dividing the bandwidth of the weight of each link on the path by taking the calculation index as another weight reference index, and comprises the following specific steps:
first, a calculation force weight model is constructed:
wherein C is br Is the total calculation force requirement; f (x) is a mapping function; alpha, beta and gamma are mapping proportionality coefficients; q 1 、q 2 、q 3 Calculating force for redundancy; j-to-m summation operation is used to sum the values of the mapping functions representing the parallel computing power available to m parallel computing chips b, where a, b, c are used to distinguish between different types of chips, m, n, p are used to distinguish between the respective number of different types of chips, i, j, k are records of different typesThe counting mark variable of the chip has significance only in the summation formula; calculation force weight = 100- (idle link calculation force +.current link calculation force +.100)
Secondly, selecting the maximum value of the calculation force weights of all links on the path as the calculation force weight of the path;
and finally, comprehensively judging the optimal weight path according to the algorithm weight and the bandwidth weight sequence.
6. The method for combining AI algorithm path optimization based on elastic communication network as set forth in claim 5, wherein: the bandwidth weight of each path is compared, and a group of paths with smaller weight are selected as preferred routing paths; the weight of at least one path in the group of paths is the minimum value in the weights of all the selectable routing paths; and after the weights of the rest preferred routing paths are differed from the weights of the paths, the difference is within a preset tolerance range.
7. The method for combining AI algorithm path optimization based on elastic communication network as set forth in claim 6, wherein: the Link between the elastic communication network node devices is abstracted to be Link, and the Link from the device to the terminal host is abstracted to be EdgeLink; in the available topological graph, the last kilometer edge link is not included; and adding two edge links in front of each other in the result of the optimal routing path obtained by the topological graph to form a complete path, and then issuing a path decision to the network.
8. The method for combining AI algorithm path optimization based on elastic communication network according to claim 1, wherein: the selectable routing paths comprehensively calculate the shortest path by adopting a k shortest path algorithm and a shortest disjoint path algorithm-Suurballer algorithm.
9. The method for combining AI algorithm path optimization based on elastic communication network as recited in claim 7, wherein: the k-shortest path algorithm adopts a deviated pathThe algorithm specifically comprises the following steps: first, the shortest path from the source point s to the target point t is found in the topology map G as the first path p i ＝{p i [0]，p i [1]… }; the method comprises the steps of carrying out a first treatment on the surface of the Then delete path p in turn i Each node p i [j]Delete edge (p) i [j]，p i [j+1]) Then find p in the remaining topology G i [j]The shortest path to t is then added to the candidate path set { C ij In }; wherein to avoid ring generation in the path, p i [j]The new path to t is queried on the remaining topology graph G'; finally in these candidate path sets { C ij Finding a shortest path from s to t as the next shortest path; the entire process loops in turn until the first k shortest paths are found or no new shortest paths are generated.
10. The method for combining AI algorithm path optimization based on elastic communication network as recited in claim 7, wherein: the Suurbelle algorithm calculates the shortest disjoint path from the source node to the destination node, and can calculate the working path and the protection path at the same time, and the method specifically comprises the following steps:
(1) The network topology can be abstracted as G (V, L, F), where V is the combination of nodes in the network, L is the set of links, F is the set of certain available resources, the traffic demand is expressed as R (s, d, w), (s, d) is the source-destination node pair, and w is the transmission demand for a certain resource; in graph G (V, L, F), a Dijkstra algorithm is used to calculate a shortest path tree T from source node s, the tree T containing the shortest paths from s to any node u; setting P as the shortest path from s to t;
(2) Updating each link weight in graph G with the notation w (u, v) =w (u, v) -d (u, v) +d (s, u); wherein w (u, v) represents the weight of a link (u, v) between node u and node v, the weight being related to a certain resource size of the link, the more resources the smaller the weight; d (s, v) + is the path weight of node s to node v; after updating, the link weight in the tree T is 0, and the link weight in the non-tree T is non-negative;
(3) Removing the link on the P path in G, and updating the reverse path link weight corresponding to P to 0;
(4) The shortest path P2 from the source node s to the destination node t in the graph G is updated after being calculated by using the Dijkstra algorithm again;
(5) And deleting the shared link segment of P and P2 in the graph G, and combining the rest links into two paths, namely the two required separated paths.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311409864.4A CN117336226A (en) | 2023-10-27 | 2023-10-27 | Path optimization method based on combination of elastic communication network and AI algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311409864.4A CN117336226A (en) | 2023-10-27 | 2023-10-27 | Path optimization method based on combination of elastic communication network and AI algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117336226A true CN117336226A (en) | 2024-01-02 |
Family
ID=89279251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311409864.4A Pending CN117336226A (en) | 2023-10-27 | 2023-10-27 | Path optimization method based on combination of elastic communication network and AI algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117336226A (en) |
-
2023
- 2023-10-27 CN CN202311409864.4A patent/CN117336226A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10715414B2 (en) | Network communication methods and apparatus | |
| US10616074B2 (en) | System, apparatus, procedure, and computer program product for planning and simulating an internet protocol network | |
| CN108600102B (en) | Flexible data transmission system based on intelligent cooperative network | |
| US8897141B2 (en) | Network system and routing method | |
| CA2843366C (en) | Resiliency-aware hybrid design of controller-switch connectivity in a split-architecture system | |
| US7403988B1 (en) | Technique for autonomous network provisioning | |
| US8547850B2 (en) | Transport control server, network system and aggregated path setting method | |
| CN111404822B (en) | Data transmission method, device, equipment and computer readable storage medium | |
| CN111130928B (en) | Network measurement method based on in-band detection in wide area network | |
| WO2021088801A1 (en) | Routing method, network controller, system, and storage medium | |
| CN107800624B (en) | Regional protection key information optimal roundabout path selection method | |
| CN109327401B (en) | Flow scheduling method and system of SDN (software defined network) | |
| CN108965013B (en) | System and method for rapidly opening IP and optical network service | |
| CN114448899A (en) | Method for balancing network load of data center | |
| CN104917677A (en) | Data stream forwarding control method and system | |
| CN117336226A (en) | Path optimization method based on combination of elastic communication network and AI algorithm | |
| Khoobbakht et al. | Hybrid flow-rule placement method of proactive and reactive in SDNs | |
| CN114157595B (en) | Communication system, data processing method and related equipment | |
| CN114039929A (en) | Method and device for scheduling service traffic | |
| CN113328950A (en) | SDN routing system construction method based on tree structure | |
| Osukhivska et al. | Method for Estimating the Convergence Parameters of Dynamic Routing Protocols in Computer Networks | |
| CN113595809A (en) | Flow monitoring method, related equipment and system | |
| CN112260875B (en) | Packet transport network index optimization method and system | |
| CN112448890B (en) | Path determining method, device and storage medium | |
| WO2023016539A1 (en) | Network path processing method and apparatus, storage medium, and electronic device |
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 |