Design And Implementation Of Multicast Measurement System And Analysis Of Relative Network Behavior

Multicast is one of the data delivery mechanisms in the current Internet. It is different from the well-known Unicast. Forwarding the data copy to all the network device interfaces where requests were received, it can make data transfer to all the receivers throughout the Internet with only one copy in each link. Multicast Network presented here is a set of hosts and network devices which can support multicast capability. It relies on the data delivering ability of underlying physical infrastructure, but has its own topology and information forwarding rules. New challenges in Network Management and Network Behavior Research are brought by its emergence and rapid growth, and Measurement and Analysis of Multicast Network Activities is the essentially important first step to the all of researches. The thesis studies on the characterization of IP Multicast Netowrk Activities and its results can be used in the following ongoing work about Multicast Network Management and Modelling.First of all, measurement tools and large research projects about Multicast Network are surveyed and analyzed, and measurement requirements to Multicast Network are given based on the former conclusions. The measurement system was designed and implemented , including software architecture, data interface among modules, communication mechanism, and deployment of hardware. Data Analysis module is the key point to the whole system, and Packet Classification algorithm was introduced and enhanced to solve the dynamic data aggregation issues. Data collection, data management, and data presentation modules are also discussed.After introducing design and implementation of the measurement system, the thesis presents the results from measurement and analysis to the real IP Multicast Network. On one hand, the system collects traffic information in several data links to analyze multicast traffic character; on the other hand, the system collecting multicast router's information to find out the anomaly behavior of Multicast Network. We can draw three conclusions from the analyzed results: there are still few multicast applications in our large-scale Education and Research Multicast Network at present whereas the network capacity can handle much; the percentage of multicast traffic can be ignored compared with unicast traffic in the inter-domain network traffic; network anomaly can be amplified in a multicast environment which cannot be imagined in the current uni-cast network; and the management model should be paid more attention for the future multicast deployment.The thesis discusses the classical IP multicast applications - multicast video conference as well as multicast Beacon. As a favorite tool used by administrators to measure network, Beacon can not match multicast video conference with data character and data sending mechanism. If it want to be the right tool in the next generation network, Beacon should be re-designed according to traffic characterization of multicast multimedia application.Finally, the Multicast Measurement System can monitor the abnormal behavior in IP multicast network by introducing multicast traffic flow model. It can be concluded from experiences in the network operation that most dangerous activities in IP multicast network is probing and DDoS attack.