Design of Translucent Optical WDM Networks

University Of Nebraska-Lincoln, Lincoln NE

Investigators

Abstract

The unprecedented growth in the number of networked users and the emergence of high-bandwidth end-user applications impose new challenges in the design of architectures for next-generation computer networks. Optical networks, employing Wavelength Division Multiplexing (WDM), transcend the bandwidth limitations of electronic networks by utilizing the enormous capacity of the optical fiber. Though this technology looks extremely promising, large-scale deployment of such optical networks in the future depends on a rapid convergence of communication network requirements and physical network realities. The focus of the proposed research in optical networks is to facilitate this process, by tackling several system-level challenges while acknowledging the limitations of existing devices. Wavelength Division Multiplexing of numerous multi-gigabit/sec channels is being deployed right now on existing and new fiber infrastructure in respones to explosive growth in both local and long-haul digital traffic. WDM deployment is surging because capacity may be economically provisioned by activating additional wavelength channels on existing fiber plant. However, as the fiber capacity grows, increasing strain is placed on the capacity of costly electronic switches at network nodes and access points. A reconfigurable transparent optical network seeks to provide a low-cost alternative by ensuring that the signal leaves the optical domain only at the source and destination, thereby avoiding unnecessary opto-electronic conversion. There are, however, fundamental limits imposed by physical-layer impairments that limit the length of end-to-end connections in transparent optical networks. Connections that span several nodes and large distances may acumulate severe impairments imposed by fundametal physical constraints such as fiber dispersion and nonlinearties, spontaneous emission noise, and device imperfections such as crosstalk. Since the physics of point-to-point communications performance is well understood experimentally and theoretically, it has been argued that nation-scale networks will avoid physical-layer impairments by deploying point-to-point WDM with electronic regeneration at each node. Such a network is called an opaque optical network. Unlike the case of point-to-point links and physical-layer optical devices, the analysis and simulation of complex all-optical or hybrid optical-electronic networks is still in its infancy and well-tested tools are only now being developed. The goals of this project are to propose, evaluate and study designs for the next-generation high-bandwidth WDM-based optical networks, which are compatible with the physical-layer characteristics of optical devices. In particular, the researcher plans to investigate the following research topics: Sparse Regeneration and Translucent Optical Networks: There has been a great deal of discussion regarding transparency vs. opacity in (national-scale) optical wavelength devision multiplexed (WDM) networks [11]. In [22], the researcher introduced the notion of a translucent optical network - a network which supports selective regeneration of optical signals within the network. The researcher's study showed that, for medium-scale networks translucency can help to improve the overall network performance. For larger-scale networks, where impairments introduced by fiber nonlinearities and dispersion cannot be ignored, the researcher anticipates that a higher degree of opacity may be needed to combat signal degradations, but this is an open problem for further research. The researcher also plans to investigate the effect of a few signal regenerators at select locations in a nation-wide WDM network. Routing and Wavelength Assignment with Power Considerations: Routing and wavelength assignment (RWA) is an important problem that arises in wavelength division multiplexed (WDM) optical networks. Previous studies have solved many variations of this problem under the assumption of perfect conditions regarding the power of a signal. The researcher propose to investigate this problem while allowing for degradation of routed signals by components such as taps, multiplexers, switching elements, fiber links, etc. The researcher plans to include novel amplifier and other device models to characterize the performance of the networks.