| Our era was characterized by large Information given by the Internet. The most difficultissue is to distinguish different application types from each other. Data carried by the Internetnot only has a large quantity but also counter with different applications, so it would bedifficult for us to recognize them. We should care more when classifying, for example:accuracy, privacy and so forth. But traditional ways: Port Based Classification, StatisticalCharacteristic Based Classification and Deep Packet Inspection methods would offend theconcerns we brought about, also have a classification rate which is not content. So our scholarconcludes a new method which was based on Transport Layer Characteristic; this methodcalled BLINC and balances those different methods and bring the classification to a pleasantlevel. My paper aims at learning BLINC and using BLINC to finish a classification system.The main work of this paper include the following aspects:1.Read a large number of references, introduces the theory of network trafficclassification, as well as domestic and foreign research now; access to relevant information atthe same time, learning several commonly used network traffic classification methods wereintroduced, their strengths and weaknesses; focuses on learning BLINC method;2. Using software engineering ideas to design the network traffic classification system,divide the classification system is into different modules, and designed them one by one.3.Using BLINC method to Implementation the network flow classification system andoptimize the execution time of the system; described in detail the process of packet capturemodule, data analysis module, net flow constitution module, and classification module.4Use of deep packet inspection (DPI) method to finish another system, the DPI systemis used on the efficiency and classification results compared with the BLINC methodclassification system.Finally, summarizing the whole the paper and future research prospects. |
PDF Full Text Request