IP Positioning Technology Based On Reference Point Optimization

Driven by the Internet tide,the IP positioning technology in today's society plays an increasingly important role.Advertisers can target their locations by users' IPs for precise ad serving,The operator can locate the location of the network fault by locating the IP so as to quickly process the network fault.Developed to now,IP positioning technology has emerged a lot of good IP positioning products.However,these products have their own limitations,such as some require a large number of users as a support,and some positioning accuracy only to the province.This thesis firstly introduces the current IP positioning technology based on network delay,network topology and other methods,and introduces the commonly used IP positioning products.Then leads the main content of the research and realization of this project,IP positioning technology based on reference point optimization.First of all,using the integrated positioning method for provincial-level positioning,and then select the appropriate benchmark for the provinces to locate and network measurement of the reference point,using RF algorithm to analyze the measured data to obtain municipal-level positioning results,and finally using the adjusted cosine similarity algorithm Analysis of the difference between the city benchmark and the measured point,with the smallest difference of the reference point of the geographical position instead of the measured point position,experiments show that this method can make 97% of the measured point positioning accuracy within 5KM.Mainly carried out the following three aspects of research.(1)Using off-site,free-time,distributed measurement method to solve the network delay jitter impact on the measurement is too large,the experiment shows that this measurement method can reduce the delay jitter for 37%.(2)Using RF algorithm to locate to the city level.the dimension selection problem is solved by the comparative experiment,and the conclusion is obtained that the position accuracy is highest when using the IP,time delay and the TTL value simultaneously 97.61%.(3)The adjusted cosine similarity is used to analyze the differences of the datum point and the measured point,and the weight of each dimension is analyzed through repeated experiments.Then the scoring system scoring then positioning model to determine the optimal weight setting program,to solve the similarity algorithm in the value of weight problems,experiments show that this method can make 97% of the measured point position error less than 5km.