Research On Similarity Measurement Method For Mobile Traffic Data

With the rapid development of Internet technology and the popularity of 4G communication technology,the frequency of mobile internet usage has far exceeded the notebook platform,and mobile internet has become the most important entrance to the Internet.Due to the portability of mobile devices,people can have a variety of entertainment activities anytime and anywhere,such as socialize,shop online,play games,watch HD videos and so on.At the same time,office is increasingly dependent on mobile devices.As a result,people generate massive amounts of mobile traffic data every day,and we need to analyze the mobile traffic data to obtain very valuable information.In the process of analyzing and mining mobile traffic data,the similarity research of mobile traffic data is an important premise,and the similarity measure will affect the results of subsequent analysis to a large extent.So,improving the accuracy of the similarity measurement between mobile traffic data is a basic and important task for data analysis.Therefore,this paper conducts related research on the similarity measurement method of mobile traffic data.The main research contents are as follows:(1)This paper proposed a new mobile traffic data similarity measurement method based on dynamic time warping algorithm.In the method for calculating the similarity of mobile traffic data,the traditional Euclidean distance can only calculate the mobile traffic of the same length,which greatly limits the application of this method.The dynamic time warping algorithm can linearly align the non-equal length mobile traffic data in the process of calculating the mobile traffic data,so that the similarity between the non-equal length mobile traffic data can be calculated,and the algorithm easily leads to the path in the calculation process.The over-fitting is introduced,sot the derivative dynamic time warping algorithm is introduced to optimize,and a similarity measure method based on dynamic time warping is proposed.Through experiments,the improved method has a certain precision improvement compared with other methods.(2)Through the research on the similarity measurement method of mobile traffic data,this paper also analyzes its application scenarios.Based on the characteristics of mobile traffic data,combined with the clustering results in the urban real ground data and similarity measure process,the functional area of the city is divided by selecting the cluster's cut-off value.In addition,when analyzing the time domain and frequency domain information of the mobile traffic data,it is possible to detect abnormal activities that may exist in the city.By dividing the city into functional area and detecting abnormal activities,it will be more conducive to people's life and urban development.(3)According to the similarity research and application of mobile traffic data,a mobile traffic data analysis system is designed.By introducing mobile traffic data,the similarity analysis can be studied,and the functions of urban functional area division and abnormal activity detection can be performed.At the same time,the user can also customize the algorithm of the similarity measurement,and by submitting the method to implement the file,the comparison between different methods can be realized,so that the accuracy of the method can be continuously improved.