CN114513449A

CN114513449A - Method and system for optimizing routing in domain

Info

Publication number: CN114513449A
Application number: CN202210168998.0A
Authority: CN
Inventors: 王军力; 王牧; 林丹清; 田丽
Original assignee: Beijing Huicheng Information Technology Co ltd
Current assignee: Beijing Huicheng Information Technology Co ltd
Priority date: 2022-02-23
Filing date: 2022-02-23
Publication date: 2022-05-17
Anticipated expiration: 2042-02-23

Abstract

The invention relates to a method and a system for optimizing routing selection in a domain, wherein a routing table of each router node is utilized to establish a related routing backtracking table, the table is used as a data mining object, an Apriori algorithm is adopted to mine trunk routing nodes frequently appearing in routing paths layer by layer, a trunk link formed by all the trunk routing nodes is a structure center of an intra-domain network, and data in a network autonomous system is shunted through the trunk link, so that the link bandwidth and the data throughput are improved, and the occurrence frequency of congestion problems in busy paths is reduced; and establishing a direct or indirect strong connection relation for all the common router nodes from near to far by taking the trunk link as a center according to the adjacent distance between the rest common router nodes and the trunk router node, so that partial data traffic is distributed to the trunk link under the principle of supporting the shortest routing algorithm, and the load pressure on the narrow-band path is further reduced.

Description

Method and system for optimizing routing in domain

Technical Field

The invention relates to the technical field of network communication, in particular to an intra-domain routing optimization method and system under an improved RIP protocol framework.

Background

With the rapid development of network technology, the dependence on the network is stronger and stronger, the network scale increases exponentially, and the network traffic data expands rapidly. In order to facilitate the maintenance and security management of network resources, the granularity of various autonomous systems built in an internet framework is more and more refined and diversified, so that the work center of gravity of packet data forwarding is gradually transferred from an out-of-domain network to an in-domain network, and the route forwarding workload in each level of autonomous system is increased sharply. Therefore, most enterprise networks adopt dynamic routing protocols at present. The dynamic routing selection means that the routing selection of the node is determined by depending on the current state information of the network, and the strategy can be well adapted to the change of the network flow and the topological structure, is favorable for improving the performance of the network and is suitable for complex network environments.

Rip (routing Information protocol) is the most known one of the intra-domain routing protocols, and because it uses a distributed routing Path algorithm based on distance vectors, compared with ospf (open short Path first) using a link state algorithm, it has the characteristics of simple operation and convenient maintenance, and becomes a routing protocol widely used by the internal networks of various enterprises at present.

The "best route" defined in the distance vector routing path algorithm is based on the shortest distance to the destination address, and each router maintains a vector table by the routing information exchanged by neighboring nodes, wherein the currently known shortest distance to each destination is listed in the table, so as to reduce the data delay of the whole network topology as much as possible. However, for the modern communication technology field, the "optimal routing" does not only satisfy the minimization of the average delay of packets, but also needs to pay high attention to the flow control of network data and the reduction of congestion problems. Particularly, with the gradual realization of office automation and service automation in large and medium-sized enterprises, the number of nodes in an intra-domain network is inevitably increased, and a large number of network nodes lead to the complication of a topological structure of the network and increase the throughput of network loads. However, the distance vector routing path algorithm cannot provide data traffic control service, and cannot solve the problem of network congestion.

In addition, for an ideal routing algorithm in the traditional sense, the algorithm should be equal for all users, i.e., all nodes in the network should be allocated relatively fair bandwidth resources within the same period of time. The situation of unfairly distributing resources is more serious because the network structure is more and more huge and complicated, which is mainly reflected in that in the communication process of some router nodes at the edge of a link and other nodes far away from the link, the sent messages are often blocked on a certain routing path, and the router nodes at the central position of the network are more easily distributed to network resources far more than the edge nodes due to the advantage of the router nodes in terms of physical position, so that poor service experience is brought to users at the edge nodes, and especially unexpected consequences can be brought to an enterprise pipeline adopting a multi-point cooperative working mechanism.

Disclosure of Invention

In order to solve the above existing technical problem, the present invention provides an intra-domain routing optimization method, which specifically includes:

step 1) collecting local routing table data of all router nodes in an intra-domain network, and constructing a routing backtracking table through an index relation among routing table items in the routing table, wherein the routing backtracking table consists of backtracking table items, and backtracking paths of the routing table items are recorded in the backtracking table items;

step 2) taking a data set formed by all route backtracking tables as a transaction data set, finding multi-level frequent item sets from the transaction data set step by using an Apriori algorithm, and excavating main router nodes corresponding to the levels from the multi-level frequent item sets, wherein all the main router nodes are used for establishing a main link for packet forwarding in an intra-domain network;

step 3) taking the router nodes except the main router node as common router nodes, adopting a set priority connection rule to establish strong connection for all the common router nodes, wherein the priority connection rule is as follows: searching neighbor nodes of common router nodes, preferentially selecting a main router node or branch router nodes as connection objects according to the sequence to establish unique strong connection, wherein the branch router nodes are router nodes directly or indirectly connected with the main router node;

and 4) in the process of updating the local routing table by using a routing path algorithm of the distance vector, preferentially selecting the routing table items forwarded by the strong connection to update to the local routing table items which are sent by different neighbor nodes and have the same destination address and distance.

Further preferably, the specific process of constructing the route backtracking table in step 1) is as follows:

step 101) collecting local routing table data of all router nodes in an intra-domain network, wherein the local routing table stored by the router nodes is constructed based on RIP protocol, and the routing table items in the routing table comprise destination IP addresses, next hop IP addresses and distance field values d_mM represents a routing table entry identifier;

step 102) extracting the current routing table item T_mThe destination IP address in the network is used as an index, and whether a routing table item T with the same destination IP address exists or not is searched from a routing table of a router node where the next hop IP address is located_m-1If present, and the distance field value is d_m1, executing step 103), otherwise, deleting the current routing table entry;

step 103) continuously extracting the routing table item T_m-1The target IP address in the routing table is used as an index, and the step 102) is repeatedly executed until the router node where the target IP address is located is traversed, the IP addresses corresponding to all the router nodes participating in the operation are sequenced according to the traversal sequence, and the current routing table item T is generated_mCorresponding backtracking table item B_mSaid backtracking table entry B_mThe method comprises a backtracking path consisting of IP addresses and hop counts of the IP addresses in the backtracking path.

Further preferably, the specific process of the node mining of the trunk router in the step 2) is as follows:

step 201) using each generated backtracking table item as a transaction data, and forming a transaction data set by the backtracking table items of all router nodes;

step 202) using the IP addresses on the backtracking path as items, scanning all items of each transaction in the transaction data set, accumulating the count of each item, and forming a set L of frequent 1 item sets by the items meeting the minimum support count₁And from the set L of frequent 1 item sets₁Selecting a router node where the IP address with the highest count is located as a level 1 trunk router node;

step 203) on the set L of frequent k-1 terms_k-1The elements in (1) are arranged in descending order of support count, and self-connection is performed

If (l)_i[1]＝l_j[1])∧(l_i[2]＝l_j[2])∧…∧(l_i[k-2]＝l_j[k-2])∧(l_i[k-1]＜l_j[k-1]) Represents a set L of frequent k-1 term sets_k-1Sub-item set l in_iAnd l_jIs connectable, the sub-item set l is connected_iAnd l_jGenerating a set of candidate k terms, { l }_i[1],l_i[2],…,l_i[k-1],l_j[k-1]A set C consisting of all candidate k item sets generated by the self-join operation_kWherein k is more than or equal to 2, i, j represents the sub-item set identifier;

step 204) scanning all items of each transaction in the transaction data set, wherein the accumulation comprises C_kThe transaction count of each candidate k item set in the set L, the set of frequent k item sets consisting of the item sets satisfying the minimum support count_kTo set L_kThe elements in the system are arranged in descending order according to the number of the support meters, and an item set which simultaneously contains the IP addresses of the nodes of the 1-k-1 level trunk router is extracted in sequence and is recorded as a pre-option set P_kFrom all pre-option sets P_kMiddle extraction 2 in top order^k-1IP address corresponding to a common router node, and 2^k-1The common router nodes are used as k-level trunk router nodes;

step 205) for the sorted set L_kAnd repeatedly executing the step 203) and the step 204) to obtain k + 1-level trunk router nodes, continuing iterative operation until no more trunk router nodes with higher levels can be obtained, forming a trunk route set by the excavated trunk router nodes with different levels, and arranging the trunk router nodes in an ascending order according to the levels, wherein the levels of the trunk router nodes are inversely proportional to the priority.

Further preferably, the specific process of establishing a strong connection for the common router node in step 3) is as follows:

step 301) extracting all neighbor nodes of the current common router node by using the routing backtracking table, and selecting a connection object according to the following rules:

if one trunk router node exists in the neighbor nodes, the connection between the trunk router node and the current common router node is established as strong connection, and the current common router node is named as an x-level 1-th branch router node, wherein x represents the number of the connected trunk router nodes, and the number of the branch router nodes is inversely proportional to the priority;

if at least two trunk router nodes exist in the neighbor nodes, selecting the connection between the trunk router node with low level number and the current common router node to establish as strong connection, and naming the current common router node as the x-level 1 st branch router node;

if the trunk router node does not exist in the neighbor nodes, executing step 302);

step 302) extracting the next n-hop nodes of the common router nodes which do not establish strong connection by utilizing the route backtracking table, wherein n is more than or equal to 1, and selecting a connection object according to the following rules:

if an x-level mth branch router node exists in the next n-hop nodes, selecting the connection of the branch router node and the current common router node to establish strong connection, and naming the current common router node as the x-level nth + mth branch router node, wherein m is more than or equal to 1;

if at least two branch router nodes exist in the neighbor node, selecting the branch router node with low series and branch count to establish strong connection with the connection of the current common router node, and naming the current common router node as an x-stage n + m branch router node, wherein x and m respectively belong to the series and branch count attributes of the branch router node establishing strong connection;

if no branch router node exists in the next n-hop nodes, executing step 303);

step 303) extracting the lower n +1 hop nodes of the common router nodes which are not established with strong connection by using the routing backtracking table, and repeatedly executing the step 302) until all the common router nodes establish strong connection in the intra-domain network.

Further preferably, in the step 4), an improved RIP protocol is adopted to perform the operation of exchanging and updating the routing information between the adjacent router nodes, and the specific operation process is as follows:

receiving the routing table sent by the neighbor node X, wherein the distance value from the neighbor node X to the destination address D is D₁If the destination address D does not exist in the local routing table, adding the routing table item into the local routing table;

otherwise, the distance value d is calculated₁And the distance value D to the destination address D recorded in the local routing table₂Making a comparison if d₁+1＜d₂If yes, updating the routing table entry of the neighbor node X to the local;

otherwise, if d₁+1＝d₂Extracting a next hop node Y to a destination address D recorded in a local routing list, comparing the priority of the neighbor node X with the priority of the next hop node Y, if the priority of the neighbor node X is greater than the priority of the next hop node Y, updating the routing list item of the neighbor node X to the local, and if the priority of the neighbor node X is not greater than the priority of the next hop node Y, not updating the local routing list item;

otherwise, if d₁+1＞d₂If yes, the local routing table entry is not updated;

wherein, the priority of the common router node is less than the priority of the branch router node is less than the priority of the main router node, and for more than two branch router nodes with the same grade number: the branch router node priority with low branch count > the branch router node priority with high branch count.

Further preferably, the condition needs to be satisfied in performing the k-level backbone router node mining process: the selected k-level trunk router nodes are not adjacent to the k-1 level trunk router nodes.

It is further preferred that during the selection of a backbone router node, if there is more than 2^k-1The ordinary router nodes meeting the selection condition of the k-level trunk router nodes are further counted, and the ordinary router nodes meeting the selection condition appear in all the route backtracking tablesFrequency, and selecting the first 2 with the highest frequency of occurrence^k-1And taking the common router node as a final k-level trunk router node.

Further preferably, in the process of establishing the strong connection by the common router nodes, if there are more than two branch router nodes with the same number of stages and branch count, further counting the frequency of the branch router nodes appearing in all the route backtracking tables, and selecting the branch router node with the highest frequency of appearance to establish the strong connection with the current common router node.

Further preferably, the minimum support count is expressed as:

wherein the content of the first and second substances,

represents the average of how often all router nodes appear in the transaction data set.

The invention also provides an intra-domain routing optimization system, which specifically comprises a routing optimization server, a routing collector and router nodes;

the route optimization server is connected with a route collector which is distributed and deployed in an intra-domain network, and specifically comprises:

the route backtracking table generation module: receiving local routing table data of each router node forwarded by a routing collector, and constructing a routing backtracking table through an index relation among routing table items in the routing table, wherein the routing backtracking table consists of backtracking table items, and backtracking paths of the routing table items are recorded in the backtracking table items;

a main router node mining module: taking a data set consisting of all route backtracking tables as a transaction data set, finding multi-level frequent item sets step by step from the transaction data set by using an Apriori algorithm, and excavating main router nodes corresponding to the levels from the multi-level frequent item sets;

the strong connection definition module: strong connection is established for all common router nodes by adopting a set priority connection rule, wherein the priority connection rule is as follows: searching neighbor nodes of common router nodes, preferentially selecting a main router node and branch router nodes as connection objects according to the sequence to establish unique strong connection;

strong connection information sending module: sending the identity information of the opposite side which is established as strong connection with the common router node to the appointed common router node through a route collector, wherein the identity information comprises the IP address of the router node and the grade attribute belonging to the main router node or the branch router node;

the route collector is used for collecting a local routing table close to the router node, forwarding the local routing table to the route optimization server, and simultaneously forwarding the router node identity information with strong connection relation, and the route collector does not participate in the routing information exchange and updating service of the intra-domain network;

the router node specifically includes:

routing table sending module: according to the route backtracking request message sent by the route collector, returning local route table data to the corresponding route collector;

strong connection information receiving module: receiving the identity information of the other party with a strong connection relation with the local and storing the identity information to the local;

routing table receiving module: the router node is used for receiving routing table data sent by the adjacent router nodes;

a route updating module: based on the improved RIP protocol, the local route updating is completed in a mode that a connection object with a strong connection relation is used as a preferential selection mode of the route updating.

The improved intra-domain routing optimization method and system provided by the invention have the advantages that:

constructing a stable and reliable main link: the method comprises the steps of establishing a related routing backtracking table by utilizing a routing table of each router node, taking the table as a data mining object, mining main routing nodes frequently appearing in a routing path layer by adopting an Apriori algorithm, and taking a main link formed by all the main routing nodes as a structure center of an intra-domain network, wherein the main link bears more data flow in an autonomous system of the network compared with other branch links, and then setting a high-performance router on the main link, such as a high-end router with a high-density high-speed interface, a multi-bus multi-core structure and a large-capacity routing storage, so that the link bandwidth and the data throughput are improved, and the occurrence frequency of congestion problems in a busy path is reduced.

The routing strategy with the data distribution function comprises the following steps: establishing direct or indirect strong connection relation for all the common router nodes from near to far by taking the trunk link as the center according to the adjacent distance between the rest common router nodes and the trunk router node, improving based on a distance vector routing algorithm under an RIP protocol framework, for the routing table entries which are sent by different neighbor nodes and have the same destination address and distance, the routing table entries forwarded by the strong connection are preferentially selected to be updated to the local, under the principle of supporting the shortest routing algorithm, partial data traffic is distributed to the backbone link, the load pressure on the narrow-band path is further reduced, meanwhile, more bandwidth resources can be distributed to the edge router, the occurrence probability of unequal resource distribution events is reduced, and relatively stable network service experience is provided for general public users or the synchronous working efficiency of enterprises is improved.

The routing efficiency between the extraterritorial networks is improved: according to the established backbone link and the layer-by-layer branch network, when data exchange is carried out with an external network, the backbone router node with higher priority level can be set as the boundary router of the autonomous system, so that the common router node can smoothly send the external network interaction message to the boundary router through a strong connection path and further transmit the external network interaction message to the external network.

Drawings

Fig. 1 is a flowchart of a method for optimizing intra-domain routing according to the present invention;

FIG. 2 is a schematic diagram of a network topology in one embodiment provided by the present invention;

FIG. 3 is a schematic diagram of a strong connection established by the network topology of FIG. 2;

FIG. 4 is a schematic diagram of a network topology with strong connections constructed in another embodiment provided by the present invention;

FIG. 5 is an overall framework diagram of the intradomain routing optimization system provided by the present invention;

fig. 6 is a schematic structural diagram of an intra-domain routing optimization system provided in the present invention.

Detailed Description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the disclosure are for illustration purposes only and are not intended to limit the scope of the disclosure.

With the increasing demand of various services on the internet, the current internet has an obvious defect in the ability to guarantee the QoS (Quality of Service) of a user. In the traditional intra-domain network, due to the complex overall structure and the saturated equipment performance, the requirements of different current new services on the QoS are difficult to meet, the network flow under the emerging service is more complex than the traditional network flow, and meanwhile, frequent communication among multiple nodes in the intra-domain network increasingly causes network congestion, so that the network bandwidth is not fully utilized, the time delay among terminal users is prolonged, and the like, thereby reducing the overall service performance of the network. Therefore, how to guarantee the QoS requirements of different data flows by using the currently limited network resources has become a key issue for optimizing the internal structure of the network.

As shown in fig. 1, the intra-domain routing optimization method provided by the present invention specifically includes the following steps:

The above implementation process is specifically described below by using a specific embodiment, as shown in fig. 2, in this example, 8 router nodes R are selected₁To R₈The edges in the graph represent the physical connection relationship between two adjacent nodes, and the local routing table data of the eight router nodes is collected, as shown in table 1:

table 1 local routing table data for each node

The local routing table stored by the router node is constructed based on the RIP protocol, wherein each cell represents a routing table item and comprises a next hop IP address and hop count (distance), wherein the destination address is displayed at the head row of a column, and each row of data is provided with a row of dataIndicating a routing table stored locally with the node of the current start address, R being used directly for simplicity of illustration₁…R₈And representing the corresponding IP address, wherein the routing table items between the adjacent nodes are directly connected, the hop count is 1, and the count of each more hop is increased by 1.

Using table 1, extracting the routing data in each cell one by one, according to the destination pointed by the current routing data, looking up whether there is a routing table entry of the same destination IP address in the routing table of the router node where the next hop IP address is located, and at the same time, judging whether the distance field is the last hop distance field minus one, if no matching result can be found, deleting the routing data in the last hop cell, otherwise, continuously extracting the next hop field value in the next hop node, continuously looking up the routing table entry with the same destination IP address from the router node where the IP address recorded in the field value is located (i.e. the next two hop nodes), and repeatedly executing the above matching operation until the router node where the destination IP address is located is traversed, indicating that the routing path from the start address to the destination address is verified as a legal path.

And then, sequencing the IP addresses corresponding to all the router nodes participating in the operation in the legal path according to the traversal sequence, and generating a backtracking table item corresponding to the current routing table item, wherein the backtracking table item comprises a backtracking path consisting of the IP addresses and the hop count of each IP address in the backtracking path.

With the starting address R in Table 1₇To the destination address R₃The routing table entry is as follows: router node R₇To R in a locally stored routing table entry₃The next hop node of is R₅Hop count is 3, and then from the next hop node R₅Is searched to know the existence of the routing table to the R₃The next hop node of the routing table entry is R₄The hop count is 2, the condition meets the requirement, and the slave node R continues₄The routing table of the router can be searched to know the routing table and the destination address R₃There is a direct connection relation between them, the hop count is 1, and finally, the node R can be used₄To the destination address R₃And completing the path traversal process of the current routing table entry.

Finally, the backtracking table entry is expressed as:

where the arrows point to paths and the numbers indicate hop counts.

Similarly, the backtracking entries corresponding to all cells under the legal path are counted, for convenience of display, different backtracking entries with routing path inclusion relations are compressed, node hop counts are omitted, and the simplified routing backtracking entries of each node can be represented in a set manner as follows:

R₁{R₂→R₅→R₇,R₄→R₃,R₄→R₆→R₈}

R₂{R₁,R₄→R₃,R₄→R₆→R₈,R₅→R₇}

R₃{R₄→R₁,R₄→R₂,R₄→R₆→R₈,R₄→R₅→R₇}

R₄{R₁,R₂,R₃,R₆→R₈,R₅→R₇}

R₅{R₂→R₁,R₄→R₃,R₇,R₆→R₈}

R₆{R₄→R₁,R₄→R₃,R₄→R₂,R₅→R₇,R₈}

R₇{R₅→R₂→R₁,R₅→R₄→R₃,R₅→R₆,R₈}

R₈{R₇→R₅→R₂,R₆→R₄→R₁,R₆→R₄→R₃}

with the above-mentioned group R₁Element "R" recorded in { }₂→R₅→R₇"for example, it contains three trace-back entries:

and after the backtracking table of each node is obtained through statistics, starting to execute the mining process of the main router node. The present invention employs an Apriori algorithm that uses a priori knowledge of the nature of the frequent item set, with an iterative method called layer-by-layer search, where the k item set is used to explore the (k +1) item set.

By definition, if item set I does not meet the minimum support threshold min _ sup, then I is not frequent, i.e., P (I) < min _ sup. If item A is added to item set I, the resulting item set (i.e., I @ A) is unlikely to occur more frequently than I. Thus, also IuA is not frequent, i.e., P (IuA) < min _ sup. This property belongs to a special class of properties called antimonotone (antimitotone), meaning that if a set fails the test, all its supersets also fail the same test. It is said to be inverse monotonic because the property is monotonic in the sense that testing is not passed.

In this example, the specific implementation process using Apriori algorithm is as follows:

firstly, each backtracking table entry is used as a transaction data, and any node IP address recorded on the backtracking path represents an entry. And combining the backtracking table entries of the router nodes into a transaction data set, wherein each element of the set represents one transaction data item.

Then, by scanning all the items of each transaction in the transaction data set, the transaction number of each item is accumulatedCounts of occurrences in the data set, a set L of frequent 1 item sets consisting of items satisfying a minimum support count₁And from the set L of frequent 1 item sets₁And selecting the router node with the highest counting number of the IP addresses as the level 1 trunk router node.

As shown in table 2, each row corresponds to a routing trace back table of each node, each column represents a count of a certain IP address appearing in the routing trace back table of each node, and all the counts are summarized to represent a total count of the current IP address appearing in the transaction data set. Wherein, the router node R₄The count of (2) is the highest, and the node is taken as a level 1 trunk router node. To facilitate the subsequent data statistics, the minimum support count set in this embodiment is 7, that is, all the router nodes are merged into the set L₁。

TABLE 2 count of occurrences in transaction dataset

Will set L₁The elements in (a) are arranged in descending order of the support count, and are represented as:

L₁＝{R₄,R₅,R₆,R₂,R₇,R₁,R₃,R₈}

due to R₁,R₃,R₈If the counts are equal, the three nodes may be sorted by sequence number.

For the above sorted set L₁Performing self-join operations

The self-connection process comprises the following steps: set L for a given set of frequent k-1 terms_k-1If (l)_i[1]＝l_j[1])∧(l_i[2]＝l_j[2])∧…∧(l_i[k-2]＝l_j[k-2])∧(l_i[k-1]＜l_j[k-1]) Represents a set L of frequent k-1 term sets_k-1Sub-item set l in_iAnd l_jIs connectableThen connect the child item set l_iAnd l_jGenerating a set of candidate k terms, { l }_i[1],l_i[2],…,l_i[k-1],l_j[k-1]A set C consisting of all candidate k item sets generated by the self-join operation_kWherein k is more than or equal to 2. Candidate 2 item set composition set C obtained by arithmetic operation₂Expressed as:

set C₂The symbol "|" in (a) indicates a simultaneous occurrence relationship, and each element in the set represents a candidate 2-item set. Further scanning all items of each transaction in the transaction data set, the totality comprising C₂The transaction count for each candidate 2-item set in (1), the statistical results are shown in table 3.

Candidate 2 item set	R₄\|\|R₅	R₄\|\|R₆	R₄\|\|R₂	R₄\|\|R₇	R₄\|\|R₁	R₄\|\|R₃	R₄\|\|R₈
								Transaction counting	8	16	7	4	8	14	7
Candidate 2 item set	R₅\|\|R₆	R₅\|\|R₂	R₅\|\|R₇	R₅\|\|R₁	R₅\|\|R₃	R₅\|\|R₈	R₆\|\|R₂
								Transaction counting	5	9	14	4	4	3	3
Candidate 2 item set	R₆\|\|R₇	R₆\|\|R₁	R₆\|\|R₃	R₆\|\|R₈	R₂\|\|R₇	R₂\|\|R₁	R₂\|\|R₃
								Transaction counting	2	4	4	10	5	6	2
Candidate 2 item set	R₂\|\|R₈	R₇\|\|R₁	R₇\|\|R₃	R₇\|\|R₈	R₁\|\|R₃	R₁\|\|R₈	R₃\|\|R₈
								Transaction counting	2	2	2	4	2	2	2

TABLE 3 transaction count summarization of candidate 2 item set

From the statistical results in table 3, it can be seen that the set L of frequent 2-item sets is composed of the candidate 2-item sets of not less than the minimum support degree 7₂Comprises the following steps: { R₄||R₆,R₄||R₃,R₅||R₇,R₆||R₈,R₅||R₂,R₄||R₁,R₄||R₅,R₄||R₂,R₄||R₈}

The above-mentioned set L₂The method is characterized in that the element support counts are arranged in a descending order, and then an item set which simultaneously contains the IP addresses of the nodes of the 1-level backbone router is extracted in sequence and is recorded as a pre-option set P₂Including router node R₄The item sets of (a) are respectively: r₄||R₆、R₄||R₃、R₄||R₁、R₄||R₅、R₄||R₂、R₄||R₈From all pre-option sets P₂Middle extraction 2 in top order¹IP address corresponding to each common router node, and using the 2 common router nodes as 2-level main router nodes, namely R₆And R₃。

In the same way, continue to pair the set L₂Performing self-join operations

All the generated candidate 3 item sets form a set C₃Expressed as:

rescanning all entries of transactions in a transaction dataset, the running total comprising C₃The transaction count for each candidate 3-item set in (1), and the statistical results are shown in table 4.

Candidate 3 item set	R₄\|\|R₆\|\|R₃	R₄\|\|R₆\|\|R₈	R₄\|\|R₆\|\|R₁	R₄\|\|R₆\|\|R₅	R₄\|\|R₆\|\|R₂	R₄\|\|R₃\|\|R₁
							Transaction counting	4	4	4	0	3	2
Candidate 3 item set	R₄\|\|R₃\|\|R₅	R₄\|\|R₃\|\|R₂	R₄\|\|R₃\|\|R₈	R₅\|\|R₇\|\|R₂	R₅\|\|R₄\|\|R₇	R₅\|\|R₂\|\|R₄
							Transaction counting	4	2	2	5	4	0
Candidate 3 item set	R₄\|\|R₁\|\|R₅	R₄\|\|R₁\|\|R₂	R₄\|\|R₁\|\|R₈	R₄\|\|R₅\|\|R₈	R₄\|\|R₂\|\|R₈
							Transaction counting	0	0	2	0	1

TABLE 4 transaction count summarization of candidate 3-item set

From the statistical results in table 4, it can be seen that no candidate 3 item set satisfying the minimum support count appears, so far, level 1 trunk router node R on the link is obtained₄And 2-level trunk router node R₆、R₃And forming a backbone link between the three nodes, wherein node R₄Has higher priority than the nodeR₆And R₃The priority is a preferred mode for establishing a strong connection relationship or a route updating rule when a plurality of optional nodes under the same condition exist. In addition, it is also intuitive from the network topology shown in fig. 2 that node R is compared to other nodes₄Is located at the center of the network structure, and a node R located at the edge of the network₃Also listed in the trunk router node through the operation rule, mainly because of the node R₃Can only pass through node R₄Communicate with other non-adjacent router nodes, so that the data message can be transmitted at the node R in the process of mutual transmission₃And a node R₄Bring a heavy load therebetween, and therefore it is necessary to raise the node R₃The bandwidth of (2) to avoid data congestion.

Except for the trunk router node, the remaining router nodes R₁、R₂、R₅、R₇、R₈Denoted as ordinary router nodes, strong connections are established for each ordinary router node. The strong connection does not build physical connection between the router nodes again, but on the basis of the original physical connection relation, one connection edge is preferably selected from adjacent connection edges to define the strong connection between the local router node and the outside, and under the same routing condition, the strong connection has higher priority than other common connection relations, namely routing information sent by the strong connection is preferentially updated to the local.

In this embodiment, a specific process of establishing a strong connection for a common router node is as follows:

step one, extracting all neighbor nodes of the current common router node by utilizing a routing backtracking table, and selecting a connection object according to the following rules:

if one main router node exists in the neighbor nodes, the connection between the main router node and the current common router node is established to be strong connection, the current common router node is named as an x-level 1 st branch router node, wherein x represents the number of the connected main router nodes, and the number of the branch router nodes is inversely proportional to the priority.

If at least two trunk router nodes exist in the neighbor nodes, the connection between the trunk router node with low grade number and the current common router node is established as strong connection, and the current common router node is named as an x-grade 1 st branch router node, wherein x belongs to the grade number attribute of the trunk router node establishing the strong connection.

If the neighbor node does not have a trunk router node, executing the second step;

and secondly, extracting the next n-hop nodes of the common router nodes which do not establish strong connection by using the routing backtracking table, wherein n is more than or equal to 1, and selecting a connection object according to the following rules:

if at least two branch router nodes exist in the neighbor node, selecting the branch router node with low series and branch count to establish strong connection with the connection of the current common router node, and naming the current common router node as an x-stage n + m branch router node, wherein x and n respectively belong to the series and branch count attributes of the branch router node establishing strong connection;

if no branch router node exists in the next n-hop nodes, executing a third step;

and thirdly, extracting the lower n +1 hop nodes of the common router nodes which are not established with strong connection by using the routing backtracking table, and repeatedly executing the previous step until all the common router nodes establish strong connection in the intra-domain network. I.e. every additional hop, the branch count is increased by one. When there are two different numbers of branch router nodes, the number of levels is low to establish a strong connection, and when there are two same numbers of branch router nodes, the router node with the low branch count is preferably selected because the router node with the low branch count is closer to the backbone link.

In a special case, if more than two branch router nodes with the same number of stages and branch counts exist, the frequency of the branch router nodes appearing in all the routing backtracking tables is further counted, and the branch router node with the highest appearance frequency is selected to establish the connection with the current common router node as a strong connection.

The procedure implemented for this embodiment according to the strong connection establishment rule described above is:

first, a router node adjacent to the backbone link, including node R, is selected₁、R₂、R₅And R₈. Wherein the node R₁、R₂With only level 1 trunk router node R₄Adjacent, thus in R₁And R₄、R₂And R₄Between which a strong connection is established, node R₅Simultaneously with node R₄And R₆Adjacent but due to R₆Belongs to a level 2 trunk router node, and the priority of the node is lower than that of a level 1 trunk router node R₄Thus, R is preferably selected₅And R₄Establish a strong connection between them, for node R₈Since it is only connected to level 2 trunk router node R₆Adjacent, thus in R₈And R₆The strong connection is established between the nodes, the establishment of the strong connection of all common router nodes adjacent to the backbone link is completed at the moment, and the node R is defined according to the router node definition rule₁、R₂、R₅Are all defined as level 1 branch-1 router nodes, since node R₈Is connected with a 2-level trunk router node R₆A strong connection relationship is established, so that the node R₈Is defined as a level 2 branch 1 router node.

Then, the last remaining node R is searched for by continuously pressing the next hop distance₇Which are respectively connected with the node R₅、R₈Adjacent due to node R₅Is higher in priority than the node R₈Thus in R₇And R₅A strong connection is established between them. As shown in fig. 3, a thick line indicates a strong connection relationship between two adjacent nodes. In addition, for node R on backbone link line₃And p₄、R₄And R₆And can also be established as a strong connection relationship with each other according to the above-mentioned rules.

According to the strong connection relationship shown in FIG. 3, if level 1 trunk router node R is used in the local area network₄The border router is set, all strong connections are guaranteed to have enough bandwidth, the main router nodes are required to select a high-end router with high-density high-speed interfaces, a multi-bus multi-core structure and a large-capacity routing storage function, other non-border router nodes have the most optimized routing path, the strong connections established by the nodes are used for communicating with an external network, and the data exchange efficiency of the local routing nodes and the external network can be greatly improved.

After the above strong connection establishment is completed, an intra-domain routing table update procedure may be performed. The invention still selects the traditional RIP protocol frame as the basic routing protocol and incorporates the priority selection strategy set by the method in the routing process of the RIP protocol.

The improved RIP protocol is adopted to execute the routing information exchange and update operations between adjacent router nodes, and the specific operation process is as follows:

otherwise, if d₁+1＝d₂Extracting a next hop node Y to a destination address D recorded in a local routing list, comparing the priority of the neighbor node X with the priority of the next hop node Y, if the priority of the neighbor node X is greater than the priority of the next hop node Y, updating the routing list item of the neighbor node X to the local, and if the priority of the neighbor node X is not greater than the priority of the next hop node Y, not updating the local routing;

otherwise, if d₁+1＞d₂Then not to the local roadUpdating;

With node R₂For example, it receives from the neighboring nodes R respectively₁、R₄、R₅For a destination address of R₈Assuming that the following routing data is received:

routing table sender	Next hop address	Distance between two adjacent plates	Destination address
				R₁	R₄	3	R₈
R₄	R₆	2	R₈
				R₅	R₇	2	R₈

Assume that node R was after the previous route update₂Locally reserved to destination address R₈Routing table entry of (2) routing node R₅Providing, then executing the following process at the time of the update:

due to the node R₁The routing distance 3+1 of the transmission is greater than the node R₂If the route distance is 3, the route data is directly deleted;

node R₅To R₈Is unchanged, node R is thus not connected to the network₅The sent routing data is the same as that sent in the previous time, and the node R is further judged₄The transmitted routing data has the transmitted routing distance of 2+1 and the node R₂If the locally stored routing distances are the same, two next-hop nodes R need to be compared₄And R₅Priority of, apparently node R₄Is higher than node R₅And node R, and₂and node R₄Have a strong connection relationship between them, so that the node R is connected₄Sent routing data replacement node R₂Routing data originally stored locally. It should be noted that if node R is a node₅The routing data is sent from the node R₂Received after the route data replacement occurs locally, and then also due to the node R₅Is less than node R₄Then node R₂Local direct delete node R₅The original route information is maintained for the route data of (1).

For the above routing optimization method, the present invention further provides another embodiment, which is different from the foregoing embodiment in that: the minimum support count set in this embodiment is adjustable. The minimum support count set when mining frequent 1, 2 item sets is still 7, and the minimum support count set when mining frequent 3 item sets is 4. The network topology structure of the mined object still continues to use the previous embodiment, and then the mining process and result for the level 1 and level 2 trunk router nodes are the same as the previous embodiment, see the previous implementation process, and are not described herein again. The following mainly explains the mining process of frequent 3 sets and 3 levels of trunk router nodes:

from the transaction count results shown in table 4, the candidate 3-item sets satisfying the minimum support count of 4 constitute a set L of frequent 3-item sets₃Comprises the following steps:

L₃＝{R₅||R₇||R₂,R₄||R₅||R₃,R₄||R₆||R₈,R₄||R₆||R₁,R₄||R₃||R₅,R₅||R₄||R₇}

the above-mentioned set L₃Is arranged in descending order of element support count, because R₅||R₇||R₂And R₅||R₄||R₇Does not contain a level 2 trunk router node R₃Or R₆Then it is deleted and R₄||R₆||R₃Also, multiple level 2 trunk router nodes are included, and are therefore also deleted. Sequentially extracting nodes R simultaneously containing 1-level trunk router₄And level 2 trunk router node R₃Or R₆Is recorded as a set of pre-options P₃The set of eligible items includes: r is₄||R₆||R₈、R₄||R₆||R₁、R₄||R₃||R₅。

According to the node mining strategy of the trunk router provided by the invention, all the pre-option sets R should be selected₃ Middle extraction 2 in top order²And the common router node is used as a 3-level trunk router node. Considering the limited network nodes selected in this embodiment, only one node is selected as the 3-level backbone router node, wherein the node meeting the selection condition is R₈、R₁、R₅. Need to further count the node R₁、R₅、R₈The frequency of occurrence in the transaction data set can be known by referring to the counting result of the statistics in Table 2, and the node R₅Is the highest, so node R is assigned₅As a 3-level backbone router node.

For set L₃Performing self-join operations

All the generated candidate 4 item sets form a set C₄Expressed as:

{R₅||R₄||R₇||R₂,R₄||R₆||R₃||R₈,R₄||R₆||R₃||R₁,R₄||R₆||R₃||R₅,R₄||R₆||R₈||R₁,R₄||R₃||R₅||R₇}

rescanning all entries of transactions in a transaction dataset, the running total comprising C₄The transaction count for each candidate 4-item set in table 5 is shown as a statistical result.

Candidate 4 item set	R₅\|\|R₄\|\|R₇\|\|R₂	R₄\|\|R₆\|\|R₃\|\|R₈	R₄\|\|R₆\|\|R₃\|\|R₁
				Transaction counting	0	2	0
Candidate 4 item set	R₄\|\|R₆\|\|R₃\|\|R₅	R₄\|\|R₆\|\|R₈\|\|R₁	R₄\|\|R₃\|\|R₅\|\|R₇
				Transaction counting	0	2	2

TABLE 5 transaction count summarization of candidate 4-item set

The transaction count in table 5 is 2 at the highest, and does not satisfy the requirement of the minimum support count 4 set in the previous iterative operation process, and the mining operation of the backbone router node is terminated.

For the remaining ordinary router nodes R₁、R₂、R₇、R₈A strong connection is established. As shown in fig. 4, the strong connection relationship obtained in this example is the same as that established in the previous embodiment. The difference lies in that: a router node R is added after the strong connection of each node is established₉Node R₉And node R₂、R₅、R₇Adjacent to each other. Due to the node R₅Is higher than the other two nodes, so node R is assigned₅And R₉The physical connection between them is determined to be a strong connection. As can be seen from the network topology of FIG. 4, a newly defined 3-level backbone router node R₅Can effectively shunt node R₉The data to be transmitted.

The number of internal nodes of different ad hoc networks is different, sometimes by a large order of magnitude. For the intra-domain network with a large number of router nodes and a complicated physical connection structure, the minimum support count can be further determined by the following formula:

wherein, the first and the second end of the pipe are connected with each other,

a mean value representing the frequency of occurrence of all router nodes in the transaction data set; in order to meet different network requirements, a dynamically variable minimum support count can be adopted in multiple excavation processes, and the minimum support count decreased layer by layer can obtain more effective excavation results.

In addition, in order to avoid the tendency of the trunk link mining to be localized, conditions can be set in the process of executing the node mining of the k-level trunk router: the selected k-level main router nodes are not adjacent to the k-1-level main router nodes, so that a main link formed by all levels of main router nodes can be extended to the whole network, the function of controlling most network node data streams is realized, and the situation that the data streams are too concentrated in a small part of area to obtain effective data distribution is avoided.

In order to implement the above-mentioned route selection optimization method, the present invention also provides an intra-domain route selection optimization system, which includes a route optimization server, route collectors and router nodes, such as the system overall connection relationship frame shown in fig. 5, and each route collector is used to collect node route data within the respective radiation cloud area range and is uniformly distributed in the whole network, and all the route collectors are used to route the route optimization server and perform centralized management and maintenance.

The route optimization server is connected to a route collector distributed in an intra-domain network, and as shown in fig. 6, the route optimization server specifically includes:

strong connection information sending module: sending the identity information of the opposite side which has a strong connection relation with the common router node to the appointed common router node through a route collector, wherein the identity information comprises the IP address of the router node and the grade of the main router node or the branch router node;

the router node specifically includes:

routing table receiving module: the router is used for receiving routing table data sent by adjacent router nodes;

a route updating module: and finishing local route updating by taking the connection object with strong connection relation as a preferred mode of route updating.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

1. A method for optimizing routing within a domain, the method comprising:

step 3) taking the router nodes except the main router node as common router nodes, adopting a set priority connection rule to establish strong connection for all the common router nodes, wherein the priority connection rule is as follows: searching neighbor nodes of common router nodes, preferentially selecting a main router node or branch router nodes as connection objects according to the sequence to establish unique strong connection, wherein the branch router nodes are the router nodes directly or indirectly connected with the main router node;

2. The intra-domain routing optimization method according to claim 1, wherein the specific process of constructing the route trace-back table in step 1) is as follows:

step 101) collecting local routing table data of all router nodes in an intra-domain network, wherein the local routing table stored by the router nodes is constructed based on RIP protocol, and the routing table items in the routing table comprise destination IP addresses, next hop IP addresses and distance field values d_m；

3. The intra-domain routing optimization method according to claim 1, wherein the specific process of the backbone router node mining in step 2) is:

If (l)_i[1]＝l_j[1])∧(l_i[2]＝l_j[2])∧…∧(l_i[k-2]＝l_j[k-2])∧(l_i[k-1]＜l_j[k-1]) Represents a set L of frequent k-1 term sets_k-1Sub-item set l in_iAnd l_jIs connectable, the sub-item set l is connected_iAnd l_jGenerating a set of candidate k terms, { l }_i[1],l_i[2],…,l_i[k-1],l_j[k-1]A set C consisting of all candidate k item sets generated by the self-join operation_kWherein k is more than or equal to 2;

step 205) for the sorted set L_kAnd step 203) and step 204) are repeatedly executed to obtain k + 1-level trunk router nodes, iterative operation is continued until higher-level trunk router nodes can not be obtained any more, the excavated trunk router nodes at all levels form a trunk route set, and the trunk route set is arranged in an ascending order according to levels, wherein the levels of the trunk router nodes are inversely proportional to the priority.

4. The intra-domain routing optimization method according to claim 1, wherein the specific process of establishing a strong connection for a common router node in step 3) is:

if no branch router node exists in the next n-hop nodes, executing step 303);

5. The intra-domain routing optimization method according to claim 1, wherein in step 4), an improved RIP protocol is used to perform routing information exchange and update operations between neighboring router nodes, and the specific operation process is as follows:

receiving the routing table sent by the neighbor node X,for a distance value D from the neighbor node X to the destination address D₁If the destination address D does not exist in the local routing table, adding the routing table item into the local routing table;

wherein, the node priority of the common router < the node priority of the branch router < the node priority of the trunk router, for more than two branch router nodes with the same progression: the branch router node priority with low branch count > the branch router node priority with high branch count.

6. The intra-domain routing optimization method according to claim 1, wherein a condition is required to be satisfied in performing k-level backbone router node mining: the selected k-level trunk router nodes are not adjacent to the k-1 level trunk router nodes.

7. The intradomain routing optimization method of claim 1, wherein during the selection of the backbone router nodes, if there are more than 2, then^k-1The ordinary router nodes meeting the selection condition of the k-level trunk router nodes further count the frequency of the ordinary router nodes meeting the condition in all the route backtracking tables, and select the top 2 with the highest frequency^k-1And taking the common router node as a final k-level trunk router node.

8. The intra-domain routing optimization method according to claim 1, wherein in the process of establishing the strong connection by the common router nodes, if there are more than two branch router nodes having the same number of stages and branch counts, the frequency of occurrence of the branch router nodes in all routing backtracking tables is further counted, and the connection between the branch router node with the highest frequency of occurrence and the current common router node is selected to be established as the strong connection.

9. The intradomain routing optimization method of claim 1, wherein the minimum support count is expressed as:

wherein the content of the first and second substances,

10. An intra-domain routing optimization system is characterized by comprising a routing optimization server, a routing collector and router nodes;

the router node specifically includes:

a route updating module: based on the improved RIP protocol, local route updating is completed by taking a connection object with a strong connection relation as a preferred selection mode of route updating.