Elimination of All-Optical Cycles in Wavelength-Routed Optical Wide Area Networks

点击翻译按钮获取中文摘要

9521249 Mukherjee A transparent (wide-area) wavelength-routed optical network may be created by using wavelength cross-connect switches connected together by fiber to from an arbitrary mesh structure. The network is accessed through electronic stations that are attached to some of these cross-connects. Such a wavelength-routed optical network is being studied and demonstrated as a part of ARPA-sponsored programs at Bellcore (ONTC and MONET). The wavelength cross-connect switches may be implemented using acousto-optic filters (AOTFs) or WDM multiplexer/demultiplexer-based technologies. The wavelength cross-connects may or may not include a wavelength-translation functionally. These wavelength cross-connect switches have the property that they may configure themselves into unspecified states. Each input port of a switch is always connected to some output port of the switch whether or not such a connection is required for the purpose of information transfer. Due to the presence of these unspecified states, there exists the possibility of setting up unintended all-optical cycles in the network (viz. a loop with no terminating electronics in it). IF such a cycle contains amplifiers (e.g., Erbium-Doped Fiber Amplifiers (EDFAs), there is a possibility that the net loop gain is greater than the loop loss. The amplified spontaneous emission (ASE) noise from amplifiers can build up in such a feedback loop to saturate the amplifiers and result in oscillations of the ASE noise in the loop. Such all-optical as defined above (and hereafter referred to as "white" cycles) must be eliminated from an optical network in order for the network to perform any useful operation. Furthermore, for the realistic case in which the wavelength cross-connects result in signal cross-talk, there is a possibility of having closed cycles with oscillating cross-talk signals. We will investigate both the network-routing layer and the physical layer solutions to this problem. At the network-routing layer, , w e will examine algorithms that set up new transparent optical connections upon request avoiding the creation of such cycles in the network. These algorithms will attempt to find a route for a connection and then (in a post-processing fashion) configure switches such that white cycles that might get created would automatically get eliminated. Furthermore, for the more realistic case in which the cross-connects result in cross-talk, we propose call-set-up algorithms that minimize the possibility of cross-talk cycles. At the physical layer, we will study various device-related issues that arise in transparent wavelength-routed optical networks. The physical layer study will directly influence the network-routing layer work stated above. Specifically, we propose (1) to provide device characterization and usable computer models of present and future optical devices such as optical amplifiers, optical wavelength cross-connect switches (with and without wavelength translation), and fiber links including connector models to network theorist so that "ideal" network models may be upgraded to "realistic" network models; (2) to explore potential solutions in the physical layer to problems identified by network demonstrations and network analysis; and (3) to identify viable alternative optical devices, particularly amplifiers and switches, whose physical properties will directly benefit the implementation of transparent optical networks. We will feed back the results of this work so as to directly benefit the ARPA-sponsored ONTC (Bellcore, BNR/NT, Columbia University, Hughes, LLNL, Rockwell, UTRC) and MONET (AT&T, Bellcore, Pacific Telesis, Bell Atlantic, Bellsouth) consortia. ***