Research On IP City-level Geolocation Based On Network Topology Clustering

With the rapid development of the Internet,network applications is playing an increasingly important role in people's daily life and work,IP geolocation is one of the indispensable technology.In the age of the big data,many traditional areas have begun to gradually carry out the research of the big data applications,but IP geolocation still rely on network measurement to locate the target IP,there is a bottleneck on efficiency and accuracy of IP geolocation,how to use the method of big data or under the big data conditions to research on IP geolocation technology has become a new challenge.Complex network and clustering technology is a typical scenario of the big data,and it is highly coupled with IP positioning.In this paper,the IP city-level geolocation problem as the starting point,by analyzing the problems of the IP city-level geolocation and the characteristics of the data,considering the use of complex networks and clustering technology to carry out the study of the IP city-level geolocation.This paper studies the IP city-level geolocation method based on community discovery,feature clustering and IP geolocation prototyping system.The main work of this paper is as follows:1.Proposes an IP city-level geolocation algorithm based on community discovery,to solve the problem that the existing IP geolocation method is not accurate.Based on the existing rulebased network topology clustering algorithm,this method introduces the community discovery algorithm in complex network.Then,it analyzes the existence of strong correlation between network topology and metropolitan area network in the network topology.Namely,the two IP of the same community belong to the same city in real network.Under the idea,through the construction of the network topology and the separation of the backbone network and nonbackbone network.Respectively,in the backbone network and non-backbone network to achieve community detection based on the optimal modularity and obtain the optimal community division.On this basis,carry out the community vote to geolocation the target IP.The experimental results of location and target IP data in seven cities in Henan Province show that the algorithm is more accurate than original algorithm in IP city-level geolocation.2.Proposes an IP city-level geolocation algorithm based on feature clustering,to solve the problem that the current IP geolocation efficiency is low and the utilization rate of historical data is not high.The existing IP geolocation method is to locating the geographical position of the target IP after getting the data of the network measurement and other parameters such as delay and hop count.In this condition,the value of delay and hops and other parameters is a one-time,that is,after this calculation can not be reused,wasting the potential value.The method proposed in this paper regards the parameters such as delay and hop count as features,the calculated geographical location is regarded as annotation,and use these features and annotation training to obtain a classifier,the original calculation problem is converted into a machine learning classification problem,making full use of its potential value,improving the accuracy and efficiency of IP citylevel geolocation.3.Designed and implemented an IP city-level geolocation prototype system based on network topology clustering.The system mainly includes the network measurement and data acquisition module,the city-level IP geolocation module and the visual module,integrating IP city-level geolocation algorithm based on community discovery and IP city-level geolocation algorithm based on feature clustering,and encapsulates highly accuracy IP location services based on the Baidu Maps API.The community structure,feature similarity and IP positioning results can be displayed on the graph through the visualization module,which is convenient for the researchers to analyze and study the IP city level geolocation technology based on network topology clustering.