The Research On Automatically Identifying Android Application In Mobile Traffic

With the rapid development of mobile network and smart phones,recent years have witnessed an exponential growth of the number of mobile apps,consequently result in the explosive increasing and more complexity of the traffic in mobile network.Considering the security and management issues,network operators need to have a clear visibility into the apps running in the network.So many methods about identifying mobile apps have come into being.Traditional technology of network traffic identification based on port number,the load on application-level,the behavior of hosts or machine learning,these technologies are widely studied and developed.And they work well on Internet network traffic identification in a certain time.However,at home and abroad,the researches about mobile network traffic identification are rare.Nowdays,analyzing the special signatures of mobile apps' packet traffic is hard to extract the signatures.Through analyzing the fields in the User-Agent or the Host of the HTTP header,the accuracy is low,while the Network Profile which extracts the fingerprints of mobile apps involves a high degree of complexity.Aiming at the problems above,considering that a majority of applications carried over the HTTP/HTTPS.1.Therefore,this paper propose a novel,efficient android app identifying system according the fingerprints of the network traffic and realizing the app's automatic identifying.2.In order to handle the large volume of traffic efficiently,this paper takes advantage of Non-Negative Matrix Factorization(NMF)to perform traffic analysis and cluster similar network traffic into groups.The access patterns of individual apps that are extracted from each group can be used as fingerprints distinguishing apps from others uniquely.3.To make the fingerprints of each app validate and accurate,this paper propose a Multiple Similar Strings Clustering algorithm to optimize these fingerprints.The experimental evaluations show that the proposed approach can identify the mobile apps from random and mixed network traffic with high precision and efficient.The system can identify mobile apps' traffic automatically and effectively with less execution time?low complexity and high fault tolerance,without accessing mobile apps' private signatures.It provides a new approach for mobile app's traffic identification.