Research And Implementation Of The Network User Online Behavior Monitoring System

With the development of computer network and the popularity of network applications, the users' online demand has become more and more complex. In the paper, the study of network user online behavior monitoring system is very important. It can provide the network usage of employees for the enterprises, institutions and even government agencies unit, which can ensure the online behavior of employee visiable, raising the employee's work efficiency and quality. It is also able to better carry out the management authority unit “commonplace lazy luxury” rectification of action, improving the work style of the agency. Therefore, the research and design a set of accurate and efficient network user online behavior monitoring system have very import significance.In order to monitor the user online behavior in real time, the paper has done a large number of investigations, designed and implemented a set of relatively perfect online behavior monitoring system. Compared with the existed client network performance testing tools, the system has the advantage of wide coverage, and it can support to monitor every computer of the LAN,and carry on unified management. In this paper, the main work includes:Firstly, because the traditional behavior online system has the problem of low efficiency for processing data in the packet acquisition module, parsing module and data storage module, the paper puts forward the data producer/consumer model, and uses multi-threaded parallel processing technology to the TCP packet and UDP packet for shunt processing, to ensure the system can parallel processing data of different protocol types. In addition, the paper designs storage threads in database storage module to improve the data packets information storage in real time.Secondly, in the design of packet parsing module, the paper designs the packet forwarding module to ensure that the packet can accurate and orderly distribute processing according to TCP/IP model, and it has realized the packet analysis issued layer upon layer.Thirdly, based on the demand of getting the users' information, the system needs to analysis the user-agent string of HTTP request message. Aiming at this problem of low efficiency and poor manitainabity for ordinary string search, the paper introduces AC_BM multiple patterns matching algorithm.The system can realize the users' information in a single test processing through the construction of model tree. The algorithm has greatly improved the speed and the accuracy of the obtaining of users' information.Fourthly, combining with the characteristics of web data streams, the system has realized the measurement of the user network performance, and put forword the scheme of user searching behavior and user attention type monitoring through the search of a large number of mainstream websitefeatures.Finally, combing the actual user behavior, the paper has finished comprehensive validation for the online behavior monitoring system. The experimental results show that the system can accurately and effectively monitor the user online havavior.