Over the past decade the growth in the use and capabilities of communication networks has transformed the way we live and work. As we progress further into the information age, the reliance on networking will increase. With the expected explosive growth in data traffic, networks will be strained in terms of both transport and processing requirements. Wavelength Division Multiplexing (WDM) is emerging as a dominant technology for use in backbone and access networks. With WDM, the capacity of a fiber is significantly increased by allowing simultaneous transmission on multiple wavelengths (channels), each operating at the maximum electronic rate. Systems with between 40 and 80 wavelengths are presently being deployed for point-to-point transmission. With tens of wavelengths per fiber and transmission rates of up to 10 Gbps per wavelength, capacities that approach a Tera-bit per second can be achieved. While these WDM systems are likely to meet future transport demands, electronically processing all of this traffic at network nodes will present a significant bottleneck. Fortunately, it is not necessary to electronically process all traffic entering and leaving each node. For example, much of the traffic passing through a node is neither sourced at that node nor destined to that node. To reduce the amount of traffic that must be electronically processed at ntermediate nodes, future WDM systems will employ WDM Add/Drop multiplexers (WADMs) and cross-connects, that allow each wavelength to either be dropped and electronically processed at the node or to optically bypass the node's electronics. This project will develop mechanisms for providing optical bypass to the electronic layer thereby reducing the size and cost of electronic switches and routers in the network. A number of techniques will be explored, each of which is appropriate for different traffic scenarios. For the case of low rate stream traffic, grooming algorithms will be developed to selectively multiplex multiple low rate traffic streams onto wavelengths such that the number of wavelengths that must be processed at each node is minimized. For bursty packet traffic, topology reconfiguration algorithms will be developed to reduce the load on the electronic switches and routers via dynamic load balancing. Lastly, for large data transfers, Optical Flow Switching protocols that bypass all of the electronics in the network using all-optical end-to-end connections will be developed. The combination of the above mechanisms will reduce the size, cost and complexity of electronic switches and routers and will lead to a dramatic increase in the traffic capacity that can be supported by the Next Generation Internet (NGI).
