To meet the bandwidth demand from the traffic explosion from emerging Internet applications like IPTV, VoIP, P2P, e-business and e-healthcare as well as large-scale science applications such as high energy nuclear physics, grid computing, and remote experimentation, optical networks using wavelength-division multiplexing (WDM) technology, which divides the enormous fiber bandwidth into a large number of wavelengths, is the foremost solution. The rapid advances in dense WDM technology with hundreds of wavelengths per fiber and world-wide fiber deployment have brought about a tremendous increase in size of the photonic switches or cross-connects, the cost and difficulty associated with controlling such large cross-connects. This project investigates developing a multi-granular switching framework to reduce the complexity, cost, and size of both electronic and optical switches. The key efforts involve the investigation of reliable waveband switching, multi-granular services, and related theoretical modeling. In particular, the project will explore design of multi-layer and single-layer photonic cross-connect architectures, optimal wavelength grouping (or wavebanding), waveband protection/restoration schemes, and reliable dynamic provisioning of multi-granular services. The project seeks to yield a fundamental understanding of multi-granular optical networks.
Broader Impacts:
While the results from this research will advance the state-of-the-art knowledge in wavelength-division multiplexing (WDM) networks, they may also be extended to other networks (e.g., with time or code division multiplexing technology) providing multi-granular switching. The proposed project will develop and transfer technology to stimulate the optical networking industry. By involving both undergraduate and graduate students in the research, incorporating research agenda into both undergraduate and graduate lectures and course projects, and disseminating findings at technical conferences and journals, the project will also help train the future scientists and engineers in high demand fields related to optical networking. The course materials developed in this project will establish a new paradigm for integrating research and education in high-speed networking.
The demand for higher bandwidth from the increasing Internet users and applications is growing at an unprecedented pace. Optical networking is the technology of choice for meeting these demands. With the wavelength-division multiplexing (WDM) technology, the enormous fiber bandwidth can be divided into a large number (>1000) of wavelengths/channels and each wavelength can support up to 100Gbps or more of Internet traffic. A large number of fibers have been widely deployed across the nation, many of which are actually not lighted for transmission, called as dark fibers. One obvious question will be why not utilizing those dark fibers to meet the current bandwidth demands. One of the major reasons for this is the challenges and cost to manage the vast number of high-capacity optical signals. In an NSF-sponsored project, computer scientists at the Georgia State University have developed techniques to reduce the size, complexity and cost associated with the managing of the ever-increasing optical switching nodes and traffic. The innovative concept is to introduce a new hierarchy or band (or wavelength/traffic grouping), which enables the network to support multi-granular traffic more efficiently. By deploying multi-granular optical cross-connect (MG-OXC), the network can group multiple wavelengths/traffic together as a band, and switch the band as a single entity (i.e., using a single port) whenever possible. For example, for a micro mirror which can be used in a switching node, it makes no difference to reflect or reroute an individual wavelength (say a yellow light beam) or a group of wavelengths (say a beam consisting of yellow, orange, green, and blue colors/lights). In this way, previously one need 4x4 switches to switch the beam consisting of four colors, now one can use a 1x1 switch to handle them as a group/band. The group explored the use of efficient algorithms and architecture design and found that the cost, complexity and size of the switching nodes can be significantly reduced. Base on the study with dynamic and static traffic patterns, the size reduction can be more than one magnitude, which actually can be translated into even more cost reduction for future high-speed optical networks in terms of both capital and operating expenditures.
