CN113395211A

CN113395211A - Routing IP positioning optimization method based on clustering idea

Info

Publication number: CN113395211A
Application number: CN202110638488.0A
Authority: CN
Inventors: 张兆心; 李宁; 孙源; 郭长勇
Original assignee: Harbin Institute of Technology Weihai
Current assignee: Harbin Institute of Technology Weihai
Priority date: 2021-06-08
Filing date: 2021-06-08
Publication date: 2021-09-14
Anticipated expiration: 2041-06-08
Also published as: CN113395211B

Abstract

The invention relates to a routing IP positioning optimization method based on a clustering idea, which solves the technical problems of poor positioning accuracy and low accuracy of a routing IP positioning database in the existing network topology, obtains routing round-trip delay information of each province and city by analyzing network detection among landmarks of each province and city, constructs landmark database data, calculates the distance of a route by utilizing a delay index, and calculates the province and city position where the routing IP is actually positioned in a traceroute by referring to the landmark database data. The invention can be widely applied to occasions for determining the geographical position information of the network entity through the IP address.

Description

Routing IP positioning optimization method based on clustering idea

Technical Field

The invention relates to the technical field of computer networks, in particular to a routing IP positioning optimization method based on a clustering idea.

Background

The IP address, which is a core component of the internet, can be used to uniquely identify each host and device. The geographical location information of the network entity is determined by the IP address, called IP positioning. Many times a route for a complete route path only belongs to two to three cities. There may be a case that when the router IP is registered, the registrar is to avoid trouble, and uniformly registers a plurality of routes in one area using the same civic address. But in actual geographic space, these routes are distributed across different cities. The phenomenon that unidentified IP is uniformly attributed to a plurality of large cluster cities exists in the IP positioning data packet.

The early IP geographical position information is collected through the information reported by the user. The information collected may be erroneous or inaccurate and take a significant amount of time to collate.

With the development of IP positioning technology, the current IP positioning methods are roughly divided into two or three categories: the IP positioning method based on network measurement, the IP positioning method based on database query and the IP positioning method based on data mining.

In the IP positioning method based on network measurement, network measurement data fluctuates, and the current IP positioning model based on network measurement is not perfect enough in consideration of the relation between the data and the distance.

The IP positioning method based on database query has small positioning error at the national level, but has unsatisfactory positioning accuracy at provincial level, city level and even street level with finer granularity.

And performing data analysis on the measurement samples at the known positions by using the IP positioning method based on data mining, and finally calculating by using a probability model to obtain the geographical position with the maximum occurrence probability of the IP to be detected. However, the accuracy rate excessively depends on the selection and parameter setting of the probability analysis model, and the universality is poor.

Disclosure of Invention

The invention provides a routing IP positioning optimization method based on a clustering idea, which is good in positioning accuracy and high in accuracy, and aims to solve the problems of poor positioning accuracy and low accuracy of a routing IP positioning database in the existing network topology.

The invention provides a routing IP positioning optimization method based on a clustering idea, which obtains routing round-trip delay information of each province and city by analyzing network detection among landmarks of each province and city, constructs landmark database data, calculates the distance of a route by using a delay index, and calculates the province and city position where a routing IP is actually positioned in a traceroute by referring to the landmark database data.

Preferably, the steps include:

1) constructing a national landmark database and a national province adjacency relation database based on the geographic position;

2) constructing a reference point candidate set based on the adjacent relation between the provinces of the source IP and the target IP and the provinces;

3) screening a reference point candidate set based on network delay;

4) extracting and comparing based on the time delay characteristics, and temporarily setting a reference point;

5) calibrating the tentative reference point based on the routing rule;

6) relatively accurate positioning based on the SLG algorithm.

Preferably, step 1) selects a large number of landmarks in provinces and cities across the country, records the IP addresses and the geographic positions of the landmarks and stores the landmarks in the national landmark database; recording each province in the country and the provinces adjacent to the province, and storing the provinces in the national province adjacency relation database.

Preferably, step 2) queries province adjacency relation according to the geographical locations of the source IP and the destination IP, and constructs all possible path sets jumping from the province n where the source IP is located to the province where the destination IP is located.

Preferably, step 3) sends a data packet to the next node from the IP whose position is determined to perform detection, and obtains the network delay; converting the time delay into a distance according to the algorithm of CBG, and eliminating non-conforming provinces; secondly, points which are too different from the route direction are excluded, a reference point which forms an angle of more than 160 degrees with the direction of the target IP is calculated, and the reference point is excluded from the candidate set; and performing primary screening on the candidate set of the reference points.

Preferably, step 4) performs delay curve acquisition from the located IP to the next IP node, extracts feature values, performs delay curve drawing on all candidate reference points, extracts feature values, and finds out the candidate point with the maximum similarity as the tentative point.

Preferably, step 5) calibrates the tentative point according to whether the tentative point and the geographical location where the destination IP is located meet the routing rule.

Preferably, step 6) uses the SLG algorithm to replace the target IP with the landmark with the minimum relative delay, so as to improve the positioning accuracy.

The method and the device locate the route IP in the traceroute path, reduce the granularity of a domestic mainstream IP location database on geographical labeling, improve the accuracy of location, perfect a location precision network topological graph, determine the location information of part of the IP through a data collection process based on a client and a landmark, and are beneficial to the follow-up confirmation of the location of the route IP by utilizing time delay; various schemes are provided for obtaining the network delay, and when the network is congested or the data is unreasonable, the obtaining delay strategy is changed in time; when the routing IP is positioned, not only the network delay information but also whether the routing path accords with the routing path selection rule or not are considered, the result accords with the routing rule better, and the path is optimized; in addition, the IP positioning accuracy can be verified or corrected by analyzing the characteristic value of the IP time delay curve and the geographical position of the previous hop and the next hop.

Drawings

FIG. 1 is a schematic flow diagram of an embodiment of the present invention;

FIG. 2 is a time delay loop range of a Harbin node in an embodiment of the present invention;

FIG. 3 is a time delay loop range of Changchun node in an embodiment of the present invention;

fig. 4 is the SLG positioning third layer algorithm.

Description of the symbols of the drawings:

1. a target IP; 2. a first router; 3. a second router; 4. a first node; 5. a first detection source; 6. a second node; 7. a second detection source; 8. a first routing path; 9. a second routing path; 10. a third routing path; 11. a fourth routing path.

Detailed Description

The present invention is further described below with reference to the drawings and examples so that those skilled in the art can easily practice the present invention.

Example 1: fig. 1 is a flow diagram of routing IP positioning according to one embodiment of the invention. The embodiment provides a method for optimizing routing IP positioning based on a clustering idea, which obtains relevant information such as route round-trip time (RTT) of each province and city by analyzing network detection among landmarks of each province and city and constructs relatively accurate and complete landmark database data. And roughly calculating the distance of the route by using indexes such as time delay and the like, and calculating and deducing the provincial and urban positions where the route IP is actually located in the traceroute path by referring to the landmark database data. The method comprises the following specific steps:

step 101, establishing a national province connection relation in a form of table printing;

in this example, the route IP jumps between adjacent provinces, and establishes a province connection relation table of the nationwide provinces, which facilitates the subsequent listing of all possible province paths.

Step 102, selecting possible province connection relations;

in this example, the route IP in the traceroute path is located, where the source IP and the destination IP are generally terminal IPs, and accurate geographical location information of the source IP and the destination IP can be collected by using information sources such as a GPS, a chirp spread spectrum technique, and a WiFi. And listing province paths which the routing path may pass through by using the province connection relation.

103, acquiring a large number of accurate points of the landmark IP and the longitude and latitude of 100% in provinces and cities of China;

in the example, the names of universities and government units in the cities of the nationwide province are recorded, the local high-probability IP addresses are kept unchanged and are located locally, then, the IP addresses appearing in the official website are crawled by using a crawler technology, and the data are used as landmarks after verification and saved.

Specifically, the accuracy and accuracy of landmark information obtained by terminal-based IP positioning and crawler are high, and a database is constructed by collecting these data. And determining the positioning of the route IP in the traceroute path according to the obtained reliable IP positioning.

Step 104, determining a province set possibly existing in the route IP according to the network delay;

in this example, the IP with confirmed geographical location information is used as an anchor point to confirm the geographical location of the next hop IP in traceroute. The network delay from the anchor point to the next hop may be obtained. The algorithm according to the modified CBG uses a conversion factor of 4c/9 to convert the delay into distance, excluding non-compliant provincial paths (the conversion factor can be expanded without suitable nodes considering that the network may be congested).

Specifically, a ping command is sent from an anchor point to the next hop to acquire network delay, but because some servers forbid ping at present, when the delay cannot be acquired through ping, the network delay of a TCP data stream or port scanning such as ZMap can be acquired from a TCP protocol layer.

105, selecting the province where the most similar landmark is located as the province of the route IP according to the similarity of the characteristic values of the time delay curves;

in this example, the remaining possible points are selected, a delay curve is obtained from the anchor point IP to the next hop IP, a characteristic value is extracted and recorded as D _ t, the delay curve is drawn for the remaining possible reference points, the characteristic values are recorded as D _1, D _2, and D _3 …, and the point with the maximum similarity to D _ t in the characteristic values is found out and used as the tentative point.

Step 106, calibrating;

in this example, steps 103-105 are repeated, and the routing rules are calibrated for the tentative point.

Step 107, searching the most suitable city with the landmark in each zip code in the province according to the characteristic value of the time delay curve and the network time delay;

in this example, step 107 is similar to step 105, and extracts the urban landmark feature values to find out the one with the highest similarity as the urban tentative point.

Step 108, finding out landmarks of the city, and further reducing the range;

in this example, the same as the previous route IP of the IP to be measured, the relative distance between the two is found by using the time delay, and a circle is drawn by using the landmark as the center of the circle and the distance as the radius, so as to further narrow the range.

Step 109, determine the IP geographic location.

Example 2:

the specific embodiment will be that the detection point of Harbin city of Heilongjiang province is used as the detection point to detect Beijing, and the traceroute which needs to be relocated is assumed to be Harbin-B-C-Beijing.

All province adjacency relation data from Heilongjiang to Beijing are obtained, as shown in Table 1.

TABLE 1

A set of all possible paths for jumping from Heilongjiang n to Beijing (taking paths passing through 3 provinces and 4 provinces as an example) is constructed as shown in Table 2.

TABLE 2

Landmark data is acquired, as shown in part in table 3.

TABLE 3

Obtaining the time delay from sending a data packet (ping, ZMap or TCP data stream, etc.) from harbin to node B and converting the time delay into distance, drawing a time delay circle, as shown in fig. 2, excluding routes containing inner Mongolia, and finding several cities closest to the edge of each candidate province near the time delay circle to obtain a candidate set, such as peony, Changchun, Jilin, and Yichun in the figure. And (3) eliminating points which are too different from the direction of the route, calculating the included angle between two vectors from Harbin to Beijing and between Islands and Harbin, calculating the included angle to be more than 160 degrees, and eliminating the included angle from the candidate set to obtain the candidate set { peony river, Changchun and Jilin }.

And selecting a candidate set, obtaining a delay curve from the IP of the Harbin to the IP of the B, extracting a characteristic value as D _ B, drawing the delay curve of the rest reference points at the same time, recording the characteristic values as D _1, D _2 and D _3, and finding out the temporary point with the maximum similarity to the D _ B in the D _1, D _2 and D _3 as the point B. The route changes to "Harbin- -Changchun- -C- -Beijing".

Calibration, selecting point C { golf, assimilation, white mountain }, as shown in fig. 3, if the similarity of the time delay curve feature value of the candidate set of points C and C is the largest, but if C selects white mountain (gilin province), the next jump cannot reach beijing, so calibration is performed, and the city golf with the largest similarity of the next province is selected as the next reference point.

When the approximate provincial granularity is reached, the next step is to perform a relatively accurate location, and the more accepted method is now the SLG algorithm. The target IP is replaced with a landmark with the smallest relative delay as shown in fig. 4.

Specifically, the SLG first narrows down the possible locations of the IP to a smaller area by a delay measurement method. As shown in fig. 4, the first probe source 5 and the second probe source 7 respectively initiate probes to the first node 4 and the second node 6 of the target IP1, join the two sub-paths that start to diverge together as relative paths, which are the first routing path 8+ the second routing path 9, the third routing path 10+ the fourth routing path 11, and finally locate the target IP to the landmark with the lowest relative path delay.

In the embodiment, the route IP is positioned through the constructed landmark database and the starting IP address and the destination address of the traceroute path, so that the positioning accuracy can be improved, and the positioning precision network topological graph is perfect; in the embodiment, through the data collection process based on the client and the landmark, the positioning information of part of the IP can be determined, and the subsequent confirmation of the positioning of the route IP by utilizing the time delay is facilitated; the embodiment provides various schemes for obtaining the network delay, and when the network is congested or the data is unreasonable, the obtaining delay strategy is changed in time; in the embodiment, when the routing IP is positioned, not only the network delay information but also whether the routing path accords with the routing path selection rule or not are considered, the result accords with the routing rule better, and the path is optimized; in addition, the IP positioning accuracy is verified or corrected by analyzing the characteristic value of the IP time delay curve and the geographical position of the previous hop and the next hop, so that the effect of achieving twice the result with half the effort is obtained.

The above description is only for the purpose of illustrating preferred embodiments of the present invention and is not to be construed as limiting the present invention, and it is apparent to those skilled in the art that various modifications and variations can be made in the present invention. All changes, equivalents, modifications and the like which come within the scope of the invention as defined by the appended claims are intended to be embraced therein.

Claims

1. A route IP positioning optimization method based on a clustering idea is characterized in that route round-trip delay information of each province and city is obtained by analyzing network detection among landmarks of each province and city, landmark database data is constructed, the distance of a route is calculated by using a delay index, and the province and city position where the route IP is actually located in a traceroute path is calculated by referring to the landmark database data.

2. The method for optimizing routing IP positioning based on clustering idea as claimed in claim 1, wherein the steps comprise:

3) screening a reference point candidate set based on network delay;

5) calibrating the tentative reference point based on the routing rule;

6) relatively accurate positioning based on the SLG algorithm.

3. The method for optimizing routing IP positioning based on clustering thought of claim 2, wherein the step 1) selects a large number of landmarks in provinces and cities all over the country, records IP addresses and geographical positions of the landmarks and stores the IP addresses and the geographical positions of the landmarks in the national landmark database; recording each province in the country and the provinces adjacent to the province, and storing the provinces in the national province adjacency relation database.

4. The method for optimizing the positioning of the routing IP based on the clustering idea as claimed in claim 2, wherein the step 2) queries province adjacency relation according to the geographical positions of the source IP and the destination IP, and constructs all possible path sets jumping from the province where the source IP is located to the province where the destination IP is located.

5. The method for optimizing routing IP positioning based on clustering idea as claimed in claim 2, wherein the step 3) sends data packet to next node from IP whose position is determined for detection to obtain network delay; converting the time delay into a distance according to the algorithm of CBG, and eliminating non-conforming provinces; secondly, points which are too different from the route direction are excluded, a reference point which forms an angle of more than 160 degrees with the direction of the target IP is calculated, and the reference point is excluded from the candidate set; and performing primary screening on the candidate set of the reference points.

6. The method for optimizing routing IP positioning based on clustering algorithm as claimed in claim 2, wherein the step 4) is to perform delay curve acquisition from the positioned IP to the next IP node, extract the feature value, perform delay curve drawing on all candidate reference points, extract the feature value, and find out the candidate point with the largest similarity as the tentative point.

7. The method for optimizing the positioning of the routing IP based on the clustering idea as claimed in claim 2, wherein in the step 5), the tentative point is calibrated according to whether the tentative point and the geographical location of the destination IP conform to the routing rule.

8. The method as claimed in claim 2, wherein the step 6) utilizes SLG algorithm to replace target IP with landmark with minimum relative delay, so as to improve positioning accuracy.