Key Technology of Service-Perceptive Traffic Management on Internet
|School||PLA Information Engineering University|
|Course||Communication and Information System|
|Keywords||Business flow identification Traffic management Early identification of ERBDPI Encryption business flow identification Network Processor|
The rapid development of the Internet and its bearer services traditional business flow identity management technology has brought great challenges and how to build a new generation of high- trusted network efficient perception and flexible management of a large number of P2P services and encryption class business the problems to be solved . This paper summarizes the business flow Identity Management Research , combined with the National 863 Program \difficulties, focus on the early identification of the business flow and flexible network processor-based management issues , the main results are as follows: ( ? ) the business flow early recognition algorithm ERBDPI a load - oriented feature detection to solve the business flow of accurate and efficient identification problem . The business stream only the start of a connection packet in ERBDPI load feature detection , you do not have to wait for the end of the connection , you can identify the type of business . Having the advantage of a simple and efficient . . Relative to the the packet sampling the traditional fair sampling method , ERBDPI lower than obtain higher classification accuracy , suitable for high-speed network traffic flows, real - time accurate identification . ( ? ) Proposed encrypted traffic stream early identification method based on traffic statistics features , to solve the encryption business flow , accurate and efficient identification difficult . The method utilizes mutual entropy traffic statistics used to identify characteristics of effective selection and classifier construction phase and the recognition phase feature weighting , to improve the recognition accuracy . Select the first few packets in each flow statistical characteristics of early identification , to improve the recognition efficiency . The experimental results show that 5 packets before each flow 4-7 characterized in encrypted traffic flow more than 80% recognition accuracy can be obtained . ( ? ) - Aware traffic management system based on network processor design to achieve both high performance and flexibility of business . Maintenance programs and management module identification module concurrent streams distinguish between business priorities and resource allocation and flexible scheduling mechanism . The test results show that the system combines wire-speed processing capabilities and differentiated service management flexibility , suitable as a remodeling business flow identity management components integrated into the next generation of high - trusted network equipment .