Atm Network Business Source Model, And Traffic Management

The main goal of this paper is to provide our understanding, proposal and guiding direction for those deputing and unsolved problems in ATM technology. It is known to all that the ATM technology is still immature, so its further spread and application has been obstructed. There are new viewpoints and technology being proposed, especially in some important basic theoretical problems, such as traffic source model, traffic management, congestion control, statistical multiplexing, protocol interoperability, etc. Thus, the valid solutions and mechanism for the unsolved problems and the balanced selection rule in these solutions and mechanism are one of the keys to the application process of ATM.In these deputing problems, the traffic source model theory is a current research hot-point. Recently, the precise analysis of a lot of computer network traffic showed that a self-similar(or fractal, long-range dependent, LRD) model is probably more realistic and adequate than the traditional Poisson-Markov-like models. As a result, the current traffic management, congestion control, statistical multiplexing mechanism based on conventional models need to be revised thoroughly or even re-designed. In the bandwidth dimensioning and queuing analysis, the conventional models are too much optimistic to the traffic burstness estimation. Besides, the combination of many traffic sources will lead to aggravate, not smooth the burstness. The self-similarity traffic brought about serious implications to statistical multiplexing, which is the basis of ATM technology.In this paper we firstly analyzed and compared several models of the variable bit-rate(VBR) video service, one of the most important traffic sources in ATM. Then we summarized some current finds and results in the area of self-similar traffic research, including mathematical models, queuing analyses, Hurst parameter computation and importance simulation(IS) methods. Based on these work, we detailed analyzed the influence of Norros effective bandwidth formula to statistical multiplexing. We proposed the rule and algorithm of traffic prediction...