Spectrum Allocation With Time-varying Traffic In Flexgrid Optical Networks

In recent years, with the dramatic increase in the use of bandwidth intensive services by new network technologies, the traditional Wavelength Division Multiplexing(WDM) system gradually cannot meet the bandwidth and flexitiy requirements of new services. In this context, flexgrid optical networks meet the requirements of diversity and large capacity for transport networks. It provides high spectrum efficiency and flexibility with small frequency granularity and flexible resource allocation scheme. In real-time communications, as the bandwidth required by a service fluctuates over time, the network algorithms should allocate spectrums flexibly to meet the demands of network traffic.This paper mainly studies the problems of spectrum allocation with time-varying traffic and its corresponding control plane, which is based on software-defined optical network technology, in flexgrid optical networks. Firstly, based on two existing schemes of directly blocking and fully reallocating spectrums for time-varing traffic, a new spectrum allocation scheme is proposed, in which the spectrum allocation is implemented by non-contiguous virtual connections. The main idea of the new scheme is: when the bandwidth requirement of a service increases, a non-contiguous spectrum block can be allocated for its increasing part, without changing already allocated spectrum slots of the service. The scheme is realized by using OMNeT++ network simulator. The simulation results show that the scheme proposed in this thesis can effectively reduce the network blocking probability,improve the network spectrum efficiency and do not interrupt any existing network connections. It is more suitable for the time-varying traffic. When transporting this complex and time-varing traffic, the control plane of the optical network has to be changed. This paper introduces the software-defined optical network(SDON), which is based on OpenFlow technology, to solve the control plane problem. Based on three-layers model of SDON, we propose a cross-layer model to use network resources more reasonably and efectively.