Based On NetFPGA Network Flow Features Extraction Technique Research

Nowadays,the Internet is affecting our lives at all times.However,with the United States of Snowden events appeared,we recognized the importance of network security technology.So,country network security and information leading group is set up.Meanwhile many colleges have done in-depth research in network security area.They developed the network traffic classification system and network intrusion detection system to ensure the stability and security of the Internet.However most of network security systems realized packet capture by software.Then the system collected and analyzed information of the flow characteristics.Finally,they monitored the condition of the network environment by machine learning algorithm.These processes will waste a lot of time,the real-time performance of the entire system is poor.So it is not suitable for high-bandwidth,high-speed network system performance requirements.So,the real time network flow feature extraction and analysis technique is becoming the focus of current research.Traditional network data of collection and analysis system is based on single packet.It will make network flow classification and network incursion detection system need to collect a large number of data features.So,it is very difficult to meet the current demand for high-speed network.However traditional network data of collection and analysis system only analyzed single packet and it cannot get packet related information.So the whole network of high-level protocol analysis is becoming more difficult.To solve above problems,network flow features extraction technique is being deeply researched.In this paper that according to the research status of network flow feature extraction technology.I designed and implemented a based on NetFPGA network flow features extraction system.Then we used this system to achieve high speed network data acquisition and analysis.Research and innovations of this paper are as follows:In this paper,we do a deeply study to the key technical of network flow feature extraction such as data package extraction,IP flows matching and network flow timeout issues.Then we propose network flow feature extraction architecture based on NetFPGA.To achieve real-time IP flow features match.we need corresponding of flow match algorithm to meet O(1)time complex degrees requirements,then we respectively use Bob Jenkins of lookup3 hash algorithm and CRC16 hash algorithm combined with double port RAM to achieve IP flow real-time features match.Then we compare the performance of two algorithms,to show that they can meet the needs of network flow real-time match.For flow timeout issues,this paper proposes an efficient flow management strategy to clear the timeout flow in the flow table storage space.Then we design dynamic management mechanism to store the flow table.