A SCALABLE MULTIWAVELENGTH MULTIHOP OPTICAL NETWORK - A PROPOSAL FOR RESEARCH ON ALL-OPTICAL NETWORKS

An architectural approach for very-high-capacity wide-area optical networks is presented, and a proposed program of research to address key system and device issues is described. This paper is a summary of a proposal submitted to DARPA for research on all-optical networks. The proposed network is based upon dense multiwavelength technology and is scalable in terms of the number of networked users, the geographical range of coverage, and the aggregate network capacity. Of paramount importance to the achievement of scalability are the notions of wavelength re-use and wavelength translation. With this approach, each user is attached to the network through a generic access station equipped with a small number of optical transmitters and receivers, each operating on a different wavelength. Using this fixed set of transceivers, a clear optical channel is established by connecting each access station to a number of other access stations through a transparent optical interconnect. This distributed optical interconnect is wavelength-selective and electronically controllable, permitting the same limited set of wavelengths to be re-used among other access stations. Furthermore, by exercising the wavelength-selective switches, the wavelength-routed connectivity between stations can be reconfigured as needed. Finally, a multihop overlay network involving wavelength translation and self-routing fast packet switches permits full connectivity, if desired, among the access stations at the individual virtual circuit level. This architecture and its innovative use of both optical and electronic technologies achieves a total independence between the number of available wavelengths and the number of nodes served, thereby introducing true scalability and modularity to optical networks for the first time. As an example of the power of this approach, it is shown in this proposal that using just eight wavelengths, such a network could in principle interconnect a population of 100 million users over a nationwide geography with an expected delay equal to that of 12 hops. If each such user were to access at a rate of 1 Gb/s, an overall pool of network capacity on the order of 10 million Terabits/s would have been created. The rationale behind the approach, the requisite systems and device technologies, and a planned testbed implementation to demonstrate overall technical feasibility of the approach are discussed.