Research On Improved Matching Algorithm Based On DPI Technology

With the gradual advancement of 5G commercialization,the high-speed communication network of the Internet is just around the corner,and the data service will become the core demand of users.The analysis and application of business data has become the key research direction of operators,but the network equipment has increased rapidly and the traffic volume has grown exponentially.The number of active users is huge,and there are many types of smart terminal applications.How to manage,apply and analyze the big data is an urgent problem for operators.DPI(Deep Packet Inspection)is a technology which can complete protocol analysis based on application layer,It is applied on network monitoring,traffic control,service identification,user behavior analysis,and firewall functions.However,the DPI system designed by domestic operators has shortcomings such as small network scale,unclear division of system function modules,large memory usage,and low recognition efficiency,which is difficult to meet the demand.In view of the current status of DPI technology,this paper studies the DPI matching algorithm and DPI system architecture.The main work is as follows:(1)Improvement of the AC algorithm.Based on the research and comparison of several typical pattern matching algorithms(BF algorithm,KMP algorithm,BM algorithm,AC algorithm,WM algorithm),the design experiments verify the superiority and problems of AC algorithm.Innovatively proposed an AC matching improved algorithm-AC_RKM algorithm,which performs hash address mapping on the pattern string in the string preprocessing stage,compares and compares the hash address of the text string with the pattern string,and deletes the hash value of the pattern string character.The text character matches the remaining character substrings.When the mismatch occurs,the optimized bad character rule is called to jump to the next state,and the maximum safe jump distance is as close as possible to the non-missing.The algorithm improves the matching efficiency.Experiments show that the AC_RKM algorithm has better performance than the original algorithm.(2)Design and test of the DPI system based on SDN technology.Aiming at the problems of small network scale,difficult system connection,serious waste of resources and low efficiency of data collection and analysis,the DPI system of the existing network proposes a unified DPI system based on SDN,and gives the detailsof the design of the system.Introduction,the system introduces SDN technology and intelligent billing technology,which can realize the unified deployment of the operator's DPI system,build a simulated network environment,complete the system identification performance test,and realize the unified management of the carrier network in the future.Provide a reference with the application.The results show that the AC_RKM algorithm effectively improves the matching efficiency of the AC algorithm,and the matching efficiency is more obvious when the matching efficiency is smaller in large-scale data.The DPI system based on SDN technology has a higher recognition rate and can identify common service types.Effectively block URL and keyword filtering,the system is advanced.