Broadband IP Metropolitan Area Networks Testing And Characteristic Analysis Of Network Aggregate Traffics

With the augmentation of users in Internet, popularization of multimedia services and continuous quick increase of information flows, the accessing method by means of traditional PSTN has more and more limitation. Consequently, building Metropolitan Area IP Network and supplying users with broadband accessing is the inevitable trend. The view has been verified by the tide of deploying Broadband IP Metropolitan Area Networks. The research on Broad IP Metropolitan Area Network testing technology, however, is very elementary. Combining the characteristics of MAN and the real experiences of testing MAN, the paper proposes a solution for testing broadband IP MAN and describes the results of testing a certain MAN. Based on the analysis of Internet traffic of the MAN, it is found that the change of this aggregate traffic follows a specific rule. Therefore, the posterior part of the paper discusses the research on the IP Data Network Aggregate Traffics.Because the IP Data Networks are in the phase of developing and evolving, its traffic model, especially its aggregate traffic model, is a controversial core question at all times. Based on the analysis of themselves collected traffic data, many literatures advised some different traffic models, which included Markov Models, Regression Models,Long-Range Dependent Traffic Models and (σ,ρ)Leaky-Bucket Model. These models described the some characteristics of network traffics in special times or environments, and had supplied efficient theoretical model with the people making research on analysis of network performance, network system design and network devices design. With the development of network technology, the emergence of new network services and the tremendous change of network users'behavior, however, it is very necessary to study the present novel characteristics of network traffics. Through the time-frequency analysis of the measuring results of the factual network, it is found out that the packet rate of aggregate traffics exhibits the periodicity, and is a non-stationary norm processes. We propose the Periodic Non-Stationary Normal Model of aggregate traffics based on measuring and analyzing results.Comparable to circuit-switching system, packet's queuing issues in IP Data Networks are very special. The variability of packet transmission exists as a result of queuing issues, which includes the variability of transmission delay and packet loss. The packet's queuing issues aren't only relative to the buffer sizes and queuing polices of network devices, but are also influenced by the arriving characteristics of data packets. According to the normal characteristic of aggregate traffics in IP Data Networks, a Normal Markov Model is proposed in order to analyze the queuing phenomena of data packets through combination with the queuing theory. We describe the building method and characteristics of the Normal Markov Model, and also give a analyzing result according to the model.