The Design And Implementation Of Network Application Controller Based On Signature

Along with the development of internet, the user scale and network application increased significantly, many new network applications occurred. These new applications give user more convenience and also there hide some threat, network quality and security attract more attention. On the other hand, there are many new network protocols, some application supports user defined the network port, some malicious software fake their flow into other normal application, the rapid increase in network speed. These new features have brought many new problems and difficulties for the identification and control of network traffic. Therefore network application filtering technology is very important now.Network application identification technology is the key to the filtering technology, this thesis focus on network application identification technology, doing research in network flow feature extraction and matching technology.Getting feature from the network traffic which generated by the application is the nature of feature extracting, it’s a procedure of getting the similarities from the network traffic, just like get the common sequence from DNA or protein sequence which is common in biology research. After doing some research on multiple sequence alignment algorithm, this thesis proposed a feature extraction algorithm based on the T-Coffee algorithm.This thesis studied the feature matching procedure of network traffic. Found that when the network traffic isn’t distributed evenly, there exists a feature matching order which cost the shortest time matching the rules. Thesis proposed a feature matching algorithm based on the network traffic classification, it choose feature matching order according to the network traffic proportion.This thesis designed and implemented a prototype of filter system. The test proves that this recognition and filtering system has a better recognition rate and higher accuracy, especially with the increase number of filtering rules, the system has a better performance.