Research On Key Issues Of Core Routers In Optical Burst Switching

IT is an inevitable trend for the telecommunication networks to be broadband, intelligent, personalized and integrated. As IP protocol becomes more and more dominant, it has been regarded as a common language to integrate the services and networks, and IP over WDM has been one of targets in the next generation of optical Internet. Optical burst switching (OBS), which combines the merits of wavelength routing and optical packet switching while avoiding their defects, can provide IP based resource assignment and switching at high speed with low demand for optical components in the all-optical networks. It takes advantages of both mature electronic control processing and high-speed optical data transmission technologies, which makes OBS a very feasible and effective optical switching technology. So OBS is a competitive switching model for the next generation of optical Internet.OBS has its own unique characteristics, such as medium switching granularity, one-way reservation and delayed reservation. It is entirely different from wavelength routing and optical packet switching. This dissertation is to research on some key problems in the core router of OBS networks to discover the differences. These problems mainly include the architecture of optical switching module of the core router, scheduling algorithms and QoS provisioning, as well as investigation into some key parameters, e.g. the wavelength number, the offset time, the burst length and the basic unit of FDL.How to determine the parameters, for instance, the burst length and offset time, and how these parameters impose their influence on network performance, are practical and important problems. Chapter 2 will answer these questions. In this chapter, firstly, wavelength number and its effect on loss performance are discussed. For a core router with complete TWC capacity, each of its output port can be modeled as a queuing system and the wavelengths of the port are servers. When the wavelength number increases, the resources available for data burst will also increase and data loss will decrease. Secondly, the offset time is investigated. The simulation results show that the absolute value of the offset time has little influence on the loss performance, but its variation has great influence on the performance. Especially, the larger the variation, the worse the loss performance becomes. Thirdly, a mathematical model for the relation between the burst length and the loss performance is proposed. It is proved that the performance goes better as the increase of burst length when no FDL is deployed in the core router. Actually, the proof is verified by the computer simulation, and can be explained from another perspective. That is, when the burst becomes very long while keeping the traffic load constant, the burst will turn into a connection and the loss probability will be nearly equal to the blocking probability of wavelength routing. Fourthly, the basic unit and stage number of FDL is discussed. The loss performance will be ameliorated, as the stage number adds. But it is stable when the stage number is very big.Chapter 3 mainly focuses on the performances and the costs of the architectures of different optical switching modules with different space switching matrixes and deployments of TWCs and...