Louisiana State University and Agricultural & Mechanical College

Other

Ahmed A El-Amawy

Martin Barry Feldman

Ramachandran Vaidyanathan

Ali Mirbod

Stefan A Pascu

A system is disclosed for implementing WDM optical networks, without blocking, and without the need for wavelength converters. The method is based on a hypercube topology for connecting the nodes, and a novel algorithm for constructing a routing tree such that blocking cannot occur even under the stringent requirement of all-to-all broadcasting. The algorithm uses shortest paths and a small number of wavelengths to satisfy the all-to-all connectivity requirement. The number of wavelengths used may, in fact, be the minimum number needed to satisfy the all-to-all connectivity requirement, using shortest paths.

Non-blocking WDM optical networks

Optical Packet Switching (OPS) is considered the most desirable switching technology for the ubiquitous optical networks that carry internet traffic. OPS could provide for great flexibility, capacity, efficiency, and bandwidth utilization that current switching strategies are not capable of providing. Despite its great appeal, OPS has been hampered by some major hurdles that prevented its practical implementation. Among such hurdles are optical buffering, optical processing/update of headers and to a lesser extent synchronization. This document introduces a novel technique for implementing packet switching in the optical domain. The new approach makes it possible to find efficient and cost effective solutions to the major problems that traditionally rendered optical packet switching (OPS) impractical. The new approach is applicable to any network topology. A complete suite of solutions to all aspects of optical packet switching that take full advantage of the basic novel approach are described. We present a complete set of solutions to the issues of buffering, header processing/update, regeneration, and synchronization among others. The new approach and affiliated solutions make it possible to implement OPS directly over WDM thus offering a very flexible optical layer capable of meeting future demand in a cost effective way.

Optical packet switching

A free space all-optical crossbar switches light from a plurality of sources onto a plurality of receivers, in any arbitrary permutation or combination (including one-to-one and many-to-one permutations). The sources and receivers may, for example, be single mode optical fibers. The polarization of the light from each source is controlled by a series of polarization control devices associated with the source so as to obtain desired angular deflections through a series of polarization-dependent angular deflectors in a first deflection unit. A lens may then direct the light from each source towards its desired receiver. An optional second deflection unit containing polarization control devices associated with individual receivers redirects the light so that it is incident normally on the receivers, an advantage if the receivers are single mode optical fibers. Alternative embodiments are described to reduce the number of optical components and to provide uninterrupted high speed data flow.

Optical crossbar switch

Optical slab waveguides are used as high-speed, high-capacity interconnects for parallel or other devices. Optical slab interconnects can connect to many more elements than can conventional electrical or fiber optic buses. A multiplexing scheme called &ldquo;mode division multiplexing&rdquo; greatly increases the number of independent channels that a single slab can support. Optical slab waveguides have a potential capacity of over one million independent channels, each channel operating at 1 GHz in a single physical medium, with each channel capable of receiving input from over 1000 ports and sustaining a load of over 1000.

Optical slab waveguide for massive, high-speed interconnects

Global synchronization of an arbitrarily large computing structure, such that the clock skew between any two communicating cells is bounded by a constant, regardless of the size of the structure. The invention uses clock nodes to perform simple processing on clock signals. The clock signal is processed in a manner similar to that in which data signals are typically processed. A cyclic, or partially cyclic, network of the clock nodes, within certain topological constraints, along with timing constraints on the function of the clock nodes, maintains a constant bound on the skew.

Arbitrarily large clock networks with constant skew bound

A method of controlling a six pulse bridge phase controlled rectifier using a microprocessor is provided by the invention herein. The method involves synchronizing an oscillator to the frequency of an ac power source and driving first and second counters with the oscillator. The first counter is used in conjunction with a phase locked loop to monitor and adjust the oscillator frequency. The count of the second counter generates an .alpha. value used in a firing angle match routine. The match routine compares the desired output voltage to the cosine of .alpha. minus an inductance factor. The match routine takes less than 20 microseconds to run allowing a new .alpha. value to be checked every 0.5 degrees for a 60 hertz power source. When a match is found, the thyristors are fired and a reload value projecting a subsequent firing in 60.degree. is calculated for the second counter.

Growing community of inventors

Baton Rouge, LA, United States of America

Ahmed A El-Amawy