|
All-optical wavelength division multiplexing (WDM) networks using wavelength routing technology are considered to be a potential and practical solution to the next generation of wide area backbone networks. The ever-increasing demand for bandwidth further enhances the application of WDM optical networking technology. The main problem involved in design and optimization of wavelength-routed optical network is lightpath routing and wavelength assignment, namely, R&WA. In previous work on the R&WA problem, the project function has been analyzed by mathematic tools, such as integrate linear programming (ILP), heuristic algorithms or probability model, under the assumption of an ideal physical layer over which transmission impairments are ignored. In practical application, however, such impairments and their influence over traffic performance should be considered seriously, and the lightpath selection should be adjusted accordingly. In this paper, we propose an integrated approach for lightpath routing, which considers both the limitations of network topology and resource and the restrictions resulted from transmission impairments through the path computed by our heuristic algorithm when performing lightpath routing. The selected paths resulted from our proposed R&WA algorithm will be evaluated in a coarse-to-fine mechanism. By analyzing the impact of the optical devices (OXC, EDFA etc.) deployed through the lightpath, an optimal path is finally determined. This paper is composed mainly in three parts. For the first part, as to a given network topology, we present the ligthpath routing result, which investigate the influence of different network parameters, based on the proposed heuristic algorithm combining graph theory. For the second part, we studied the impact of various transmission impairments over BER performance, including crosstalk and noise caused by the optical devices. For the third part we combine the results obtained from the former two parts, and evaluate the selected path in a coarse-to-fine mechanism to make an optimal selection. The problem of load balancing is also briefly discussed in our paper.
|
